2014-04-01 03:10:02 thanks for the notes BitL0G1c, planning on doing it soon :P 2014-04-01 11:10:05 is there a package with the thin provisioning utilites? 2014-04-01 11:10:10 for lvm 2014-04-01 12:47:19 anyone? the commands are thin_dump, thin_restore and thin_check 2014-04-01 13:20:44 darkfaded: did you check aports? 2014-04-01 13:23:17 sadly i'd not know where to look. i can do a grep -R over the whole tree 2014-04-01 13:23:20 will do 2014-04-01 13:26:05 am nearly finished creating a package for thin-tools 2014-04-01 13:26:45 yay! 2014-04-01 13:26:58 BitL0G1c: thank you very much 2014-04-01 13:27:11 i mean, on behalf of anyone who wants to test it :) 2014-04-01 13:38:48 BitL0G1c: creating a seperate pkg for thin tools? 2014-04-01 13:43:01 yes - seems to be a bug building 2014-04-01 13:43:16 i think it also needs to be enbaled on lvm2 2014-04-01 13:43:22 enabled 2014-04-01 13:44:00 getting a bug similar to this https://github.com/jthornber/thin-provisioning-tools/pull/5 2014-04-01 13:45:34 clandmeter: i think the actual thin prov stuff in lvm2 is on, i think i made my first tests on alpine 2014-04-01 13:46:00 i see on arch they set the configure option to internal 2014-04-01 13:46:13 without the thin_repair a full pool unfortunately doesn't mean write block but full data loss 2014-04-01 13:46:16 im not sure thats the defaul option 2014-04-01 13:46:21 the whole technology is a bit shitty 2014-04-01 13:46:25 the kernel module is there 'dm_thin_pool' 2014-04-01 13:46:51 clandmeter: what does configure option internal mean? 2014-04-01 13:47:47 darkfaded: http://sprunge.us/YBSD 2014-04-01 13:48:28 --with-thin-check=PATH thin_check tool: [[autodetect]] 2014-04-01 13:48:39 would mean to me that it's not part of the default lvm buld 2014-04-01 13:48:49 but the feature can be enabled anyway 2014-04-01 13:48:58 which is pretty high risk 2014-04-01 13:50:09 im looking at fedora 2014-04-01 13:52:19 BitL0G1c: why are these tools not included in lvm2? 2014-04-01 13:53:13 because they are still experimental most probably - I think this is their status @ Red Hat 2014-04-01 13:53:56 it's no longer a feature preview now, apparently, but it's a different project that lvm itself 2014-04-01 13:54:07 we also need --enable-dmeventd 2014-04-01 13:54:49 Add '--with-thin=internal' to your configure line. 2014-04-01 13:54:50 You should have --enable-dmeventd too and install dmeventd for automatic 2014-04-01 13:54:51 extension of nearly-full thin volumes 2014-04-01 13:59:12 raised an issue for the build bug https://github.com/jthornber/thin-provisioning-tools/issues/10 2014-04-01 14:08:43 BitL0G1c: uclibc or musl dev? 2014-04-01 14:10:05 uclibc 2014-04-01 14:11:13 it fails on musl on something different 2014-04-01 15:57:06 developer replied - fixed in 0.31 2014-04-01 16:25:33 I tried packaging thin-provisioning last year for docker 2014-04-01 16:25:39 compilation gets a bit further but still fails 2014-04-01 16:26:32 but decided to use btrfs instead 2014-04-01 16:27:26 c++/boost did exactly not give me confidence in the implementation.. 2014-04-01 16:27:40 s/exactly not/not exactly/ 2014-04-01 16:28:18 uggedal: i need to test it so i can help people fix it if they weren't as critical as you and then sit there with the pieces of their data 2014-04-01 16:28:40 so stuffing something i have backups off onto it should be fine 2014-04-01 16:28:47 s/off/of/ 2014-04-01 16:42:03 updated the giithub issue 2014-04-01 19:50:33 @clandmeter - try again using https://github.com/jthornber/thin-provisioning-tools/archive/master.zip & note the latest githuub issue to fix the man dir problem 2014-04-01 20:18:50 BitL0G1c: yes i was just trying it 2014-04-01 20:18:55 it works on musl now 2014-04-01 20:19:11 but its not really pretty 2014-04-01 20:27:53 I managed to make a thin-tools & thin-tools-doc 2014-04-01 20:32:57 are we interested in dmeventd support for lvm2? 2014-04-01 20:35:01 clandmeter: i think the relevant error messages can be issued without it, but for helpless people the autoextension could be good 2014-04-01 20:35:17 it's just so unreliable i'd personally say it doesn't matter if it's there 2014-04-01 20:35:41 otoh it's the "intended" thing to have 2014-04-01 20:36:57 the two relevant alerts from thin pools are those: 2014-04-01 20:37:05 device-mapper: thin:.*reached low water mark 2014-04-01 20:37:10 device-mapper: thin:.*no free space 2014-04-01 20:37:53 i could just test if it sends those without the dmeventd 2014-04-01 20:38:01 i think ill add it 2014-04-01 20:40:06 if you can, it'll of course be easier 2014-04-02 00:20:10 I messed up and deleted linux-grsec from the system 2014-04-02 00:20:17 linux-vanilla is installed, but won't boot 2014-04-02 00:20:44 chrooting into that install and manually trying to get the kernel installed results in an APK error reading a database 2014-04-02 00:21:04 /proc and /sys are mounted 2014-04-02 00:30:19 Xe: if linux-vanilla is installed, have you tried just adjusting your bootloader to point to the vanilla kernel? 2014-04-02 00:30:36 yes 2014-04-02 00:30:39 doesn't work 2014-04-02 00:34:45 Fixed it by adding linux-grsec to /etc/apk/world and running apk fix outside the chroot 2014-04-02 00:54:47 Can anyone think of why a 2.7.5 system can ping6 it's neighbors, but not the gateway, when the neighbors can reach the gateway and beyond without any issue? 2014-04-02 00:56:55 firewall on the gateway? 2014-04-02 00:58:30 if ipv4, I'd say possibly screwy arp table on the 2.7.5 box, but I can't remember if ip6 still uses arp like ip4 does :/ 2014-04-02 01:07:14 it doesn't use arp, and those ports are allowed on my end. Shouldn't be a firewall on the gateway at all. 2014-04-02 01:14:16 I found tcpdump very useful for troubleshooting connectivity 2014-04-02 01:30:15 busy network for that... 2014-04-02 01:48:06 yea, I think something is screwey.. running tcpdump this particular alpine box just sits there asking the all routers address who has the gateway, while every other host just asks who has the gateway address who has it... 2014-04-02 02:01:55 01:47:52.545934 IP6 2604:6080:0:4::52 > 2604:6080:0:4::1: ICMP6, neighbor solicitation, who has 2604:6080:0:4::1, length 32 2014-04-02 02:01:55 01:47:52.546307 IP6 2604:6080:0:4::1 > 2604:6080:0:4::52: ICMP6, neighbor advertisement, tgt is 2604:6080:0:4::1, length 24 2014-04-02 02:01:55 01:47:53.223592 IP6 2604:6080:0:4::51 > ff02::1:ff00:1: ICMP6, neighbor solicitation, who has 2604:6080:0:4::1, length 32 2014-04-02 02:02:00 that's weird... 2014-04-02 02:07:17 the ::1 is in the routing table according to route -A inet6 2014-04-02 02:21:16 anyone ever tried to put the modloop.squashfs into the ramdisk? 2014-04-02 09:27:11 BitL0G1c: did you build thin tools on x86? 2014-04-02 11:19:55 yes 2014-04-02 11:27:51 BitL0G1c: they dont build on uclibc x86_64, they are now in musl-edge/testing only. 2014-04-02 11:29:15 ok 2014-04-02 11:45:44 what was the error ? 2014-04-02 11:46:42 BitL0G1c, http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64.v131211-1497-g9331f94.log 2014-04-02 12:08:59 ok thx 2014-04-03 05:11:11 Hello! 2014-04-03 05:11:11 Today logged on SSH as user. Tried to make su. Entered the password and got the message "su: incorrect password". 2014-04-03 05:11:11 Through the web interface added ssh-key for root, allowed to SSH root login. Tried to change the password. Received the following message "passwd: no record of root in / etc / shadow, using / etc / passwd: Numerical result out of range" 2014-04-03 05:11:11 And the password is not changed. How can I fix this? 2014-04-03 05:18:19 The situation repeated on another machine after the upgrade package busybox 2014-04-03 05:18:49 cat /etc/alpine-release - 2.6.6 2014-04-03 07:26:54 After the upgrade the following packages: libc (0.9.33.2-r21 -> 0.9.33.2-r22), busybox (1.21.1-r0 -> 1.21.1-r1), libcrypto1.0 (1.0.1e-r4 -> 1.0.1f-r0), uclibc-utils (0.9.33.2-r21 -> 0.9.33.2-r22), 2014-04-03 07:26:55 openssh-client (6.2_p2-r1 -> 6.2_p2-r2), openssh (6.2_p2-r1 -> 6.2_p2-r2), libpthread (0.9.33.2-r21 -> 0.9.33.2-r22), net-snmp-libs (5.7.2-r0 -> 5.7.2.1-r0), net-snmp-agent-libs (5.7.2-r0 -> 5.7.2.1-r0), libssl1.0 (1.0.1e-r4 -> 1.0.1f-r0), 2014-04-03 07:26:55 openssl (1.0.1e-r4 -> 1.0.1f-r0) 2014-04-03 07:26:55 Unable to login as any user through the console. You can login via SSH as any user, but it is impossible to do "su". Please help. 2014-04-03 07:37:09 oh 2014-04-03 07:37:14 you need to make /bin/bbsuid suid root 2014-04-03 07:37:21 it probably got unset somehow 2014-04-03 07:37:49 or maybe this broke things: http://git.alpinelinux.org/cgit/aports/commit/?h=2.6-stable&id=ab88f58f005a1177790e582e1f0171cc4ee5dcce 2014-04-03 07:39:12 but i'd doubt that... 2014-04-03 07:39:13 hm 2014-04-03 07:43:29 kaniini: Sorry, I'm not good at busybox. I've done "/bin/bbsuid suid root", but got the following message "/bin/bbsuid: Use --install to install symlinks". 2014-04-03 07:46:06 su should be suid root now... 2014-04-03 07:46:21 ncopa: If I install busybox of release 2.7, this error may disappear? 2014-04-03 07:48:47 jakal: i would expect so 2014-04-03 07:49:24 please let me know if it fixes it 2014-04-03 07:49:40 i have actually heard someone else having similar problem with 2.6, but i am not 100% it is same issue 2014-04-03 07:51:12 ncopa: I'm using release 2.6.6 on Alix board in another city. If I update busybox of release 2.7, the system will boot? 2014-04-03 07:51:53 I really do not want to go to another city. ))) 2014-04-03 07:52:32 fabled: you have alix board? have you tried boot v2.7 on alix? 2014-04-03 07:52:47 jakal: i dont have alix board myself so i cannot guarantee it 2014-04-03 07:53:01 ncopa, yes, alix works on 2.7 here iirc 2014-04-03 07:53:07 however, i wonder if I should revert that busybox issue 2014-04-03 07:53:12 or at least try reproduce it... 2014-04-03 07:53:30 yes, i can confirm that i have boards running 2.7.4 2014-04-03 07:53:43 thanks 2014-04-03 07:53:56 jakal: so it should work with v2.7 2014-04-03 07:54:09 but i cannot guarantee anything :-/ 2014-04-03 07:54:51 ncopa: I am running release 2.7.5 on another board Alix. So it works. If I update the package busybox - it will not break the whole system? 2014-04-03 07:55:43 Generally I need to check on the virtual machine. ))) 2014-04-03 07:56:17 no, upgrading busybox should not break the whole system 2014-04-03 07:56:18 but 2014-04-03 07:56:20 to be sure 2014-04-03 07:56:27 you can start with: apk add busybox-static 2014-04-03 07:56:41 then if busybox actually breaks dugin the upgrade 2014-04-03 07:56:45 or libc gets deleted 2014-04-03 07:56:47 or anything 2014-04-03 07:57:04 then you have an emergency system: busybox.static ls 2014-04-03 07:57:14 also apk add apk-tools-static 2014-04-03 07:57:22 then you can apk.static 2014-04-03 07:57:53 ncopa: Thank you, I will first try on virtual machines. )) And I repeated this bug on 2 real machines and one virtual. 2014-04-03 08:00:08 In all cases, the release was 2.6.6, the latest updates have been applied. Situation emerged after a reboot. 2014-04-03 08:05:03 ok 2014-04-03 08:05:15 I'll test v2.6 in a vm here and see if i can reproduce it 2014-04-03 08:17:12 maybe I'm having a bad day or just generally I'm dumb, but is there some reason I can never get /dev/random to write more than 512 bytes to a file? I have plenty of entropy, also running haveged, would expect to be able to write alot more than 512 bytes. Is this some built in kernel restriction? 2014-04-03 08:23:44 Could it be a virtual machine 'feature'? Running 2.7 in a vmware vsphere client. 2014-04-03 08:29:16 are you sure you have lots of entropy? 2014-04-03 08:29:30 is it a run-from-tmpfs setup? 2014-04-03 08:29:46 did have indeed, cat /proc/sys/kernel/random/entropy_avail reported 16xxx 2014-04-03 08:29:54 nope, disk install 2014-04-03 08:30:16 ncopa-desktop:~$ ls -l random.out 2014-04-03 08:30:16 -rw-r--r-- 1 ncopa ncopa 96 Apr 3 10:30 random.out 2014-04-03 08:30:25 ncopa-desktop:~$ cat /dev/random > random.out 2014-04-03 08:30:28 thats what i did 2014-04-03 08:30:52 its slow indeed 2014-04-03 08:30:57 i dont have much entropy 2014-04-03 08:31:01 I used dd, tried echo too last night, but never get more than 512 2014-04-03 08:31:09 ok 2014-04-03 08:31:20 what kernel? 2014-04-03 08:31:29 it could be something in grsec kernel 2014-04-03 08:31:30 *might* be a VM thing. Sadly I don't have a physical machine to test against 2014-04-03 08:31:44 i am testing on physic machine now 2014-04-03 08:32:00 ncopa-desktop:~$ ls -l random.out 2014-04-03 08:32:00 -rw-r--r-- 1 ncopa ncopa 632 Apr 3 10:31 random.out 2014-04-03 08:32:06 i got mroe than 512 2014-04-03 08:32:10 what kernel is it? 2014-04-03 08:32:12 cool 2014-04-03 08:32:30 $ uname -a 2014-04-03 08:32:30 Linux ncopa-desktop 3.12.8-0-grsec #1-Alpine SMP Wed Jan 22 09:48:19 UTC 2014 x86_64 Linux 2014-04-03 08:32:43 3.10.33-1-grsec 2014-04-03 08:32:47 alpine smp 2014-04-03 08:32:53 ok 2014-04-03 08:32:54 (64 bit) 2014-04-03 08:33:11 could you please file a bug on bugs.alpinelinux.org about it? 2014-04-03 08:33:15 so i dont forget 2014-04-03 08:33:23 I'll run some more tests and then do that :) thanks 2014-04-03 08:40:18 interesting, seems to be dd that struggles with it 2014-04-03 08:40:46 using cat like you did, I can get MB's worth of data written 2014-04-03 08:42:48 re 2014-04-03 08:42:50 morning 2014-04-03 08:43:01 in other words, this probably isn't worth your time looking into it as a bug 2014-04-03 08:43:10 and good morning StarWarsFan 2014-04-03 08:47:55 @ncopa yeah don't worry about the bug and my small obsession with random numbers, seems to be dd, doesn't seem to like 'bs' option being too large. Works ok once lowered 2014-04-03 08:50:27 :D 2014-04-03 08:50:37 ok 2014-04-03 08:50:47 yeah dd and 512 makes sense... 2014-04-03 08:51:29 indeed. I'm going to put this one down to a 'senior moment' :S 2014-04-03 09:08:19 ok, i can confirm v2.6 is totally broke 2014-04-03 09:08:36 i just saw the random discussion in backlog. on freebsd you could give a list of device interrupts to monitor, is there any way to do that on linux? 2014-04-03 09:08:49 waiting on mouse and keyboard input only, in a vm, just doesn't help 2014-04-03 09:13:13 haveged has really helped with entropy generation :) 2014-04-03 09:13:14 darkfaded: I think this is only used to gather initial entropy 2014-04-03 09:15:56 royger: and? :) 2014-04-03 09:16:52 i just miss something to do rng config on linux. there's a few flaming whitepapers about how bad everything is, i.e. from vmware, and a lot of issues with software that really wants entropy 2014-04-03 09:17:30 i.e. weblogic either takes 20min+ to start or people cheat and don't get entropy 2014-04-03 09:19:05 $customer here runs some HSM hardware but this is just wrong[tm] since you get nice closed vendor drivers and all that 2014-04-03 09:20:04 usually entropy generation seems to rely mostly on mouse and keyboard, with smaller weightings for disk activity - I can't find the resource that stated the values right now....so yeah, you need to add more sources for entropy, whether HW RNG or other software solutions 2014-04-03 09:25:24 ginjachris: i have the same issue on esxi for generating a gnupg key. had to do it on another box. 2014-04-03 09:27:25 I've started adding haveged (http://alpinelinux.org/apk/main/x86_64/haveged) as standard to all installs now 2014-04-03 09:28:04 fills the entropy pool once you go below the write count, and takes entropy from cpu flutter. 2014-04-03 09:29:15 Just generated a 200MB file from /dev/random (i.e. therefore mostly using haveged) and tested it with ENT (http://www.fourmilab.ch/random/random.zip) and got pretty solid results 2014-04-03 09:31:17 ginjachris: thanks for the info, i'll start adding it too 2014-04-03 09:33:19 np :) I've become slightly obsessed with randomness recently! Keyboard/mouse input is usually no help to me, since 99% of my Alpine boxes are remote servers. There's a page here: https://calomel.org/entropy_random_number_generators.html that you might find interesting 2014-04-03 09:35:58 that page... this girl/guy (whoknows) writes incredibly good stuff 2014-04-03 09:38:18 yeah, really useful for alot of stuff 2014-04-03 09:38:40 and yeah, I can't work out if guy/girl lol 2014-04-03 09:41:03 hmm, that reminded me of a page with ntp howtos, that of putting gps on my raspi and that of finally sending a mail about interesting arm hw to -devel 2014-04-03 09:41:09 i'll bbl :) 2014-04-03 09:52:49 ntp how-to is another page i need to add to the wiki, justhaven't got around to it yet 2014-04-03 09:52:57 along with one on randomness of course 2014-04-03 10:30:15 ncopa: Sorry, out on the spot. Your tests have confirmed bug? I tried to upgrade 2.6 to 2.7 in a virtual machine and the bug disappeared. As far as I can tell, in release 2.7 package busybox older than 2.6 2014-04-03 10:50:02 if i'd need a /dev/sg0 (scsi generic) device, how would I go about adding it? 2014-04-03 10:50:38 hehe, modprobe sg looks good 2014-04-03 10:50:59 meh, updated kernel and didn't reboot 2014-04-03 10:51:04 why. 2014-04-03 10:51:16 daveh0003:~# modprobe sg 2014-04-03 10:51:16 modprobe: can't change directory to '3.13.7-0-grsec': No such file or directory 2014-04-03 10:51:19 :)) 2014-04-03 11:33:51 jakal: yes, i tried to fix a bug in v2.6 but has only patched bb in v2.6 2014-04-03 11:33:57 i will not push that "fix" to v2.7 2014-04-03 11:34:02 unless i find out whats wrong 2014-04-03 11:34:12 basically 2014-04-03 11:34:19 v2.6 is totally broken atm 2014-04-03 11:40:57 ncopa: Maybe this will help ... I tried to update the version 2.6.2 to version 2.7.3. Update was successful, but the bug remained. After I updated the virtual machine with 2.7.3 to 2.7.5 - bug disappeared. 2014-04-03 11:42:24 i still had my fileserver on 2.6 and updated yesterday 2014-04-03 11:42:27 this will be interesting 2014-04-03 11:43:28 darkfaded: Can try to reboot the file server and report the results? :) 2014-04-03 11:43:47 sigh yeah i just rebooted the vm host for that kernel update 2014-04-03 11:43:54 i'll check if it still booted 2014-04-03 11:44:52 hmm at least it got to mounting root 2014-04-03 11:45:18 but seems it's stuck after that 2014-04-03 11:45:30 oh 2014-04-03 11:45:33 a prompt now 2014-04-03 11:46:19 vfile:~# 2014-04-03 11:46:36 it's up. what should i check for extended brokenness? 2014-04-03 11:46:41 there is a problem in that it is impossible to log in as any user through the console. 2014-04-03 11:46:49 haha yes i had that 2014-04-03 11:46:56 i thought it's me being stupid 2014-04-03 11:47:07 oh, and on my irc box i can't su any more since 1.5 weeks 2014-04-03 11:47:26 You can login via SSH to any user, but it is impossible to make "su " 2014-04-03 11:47:29 i had determined (via timestamps) it wasn't a hack and couldn't even look at it yet 2014-04-03 11:47:32 lol gotcha 2014-04-03 11:47:47 i'm pretty happy this is all that's to it. 2014-04-03 11:48:35 i forgot to enable ssh before i rebooted my vm 2014-04-03 11:48:48 and i cannot login from tty as root 2014-04-03 11:49:11 or anything else 2014-04-03 11:49:20 seems like password auth is completely broke 2014-04-03 11:49:21 mount / chroot? 2014-04-03 11:49:28 yup, thats what i need to do 2014-04-03 11:49:32 urgh... 2014-04-03 11:49:53 what was there, ah right, if you wanna apk upgrade make sure you also mount /run and make a "lock" folder in it 2014-04-03 11:50:06 Yes, if SSH is disabled, you can not log in to the system in any way in another city. 2014-04-03 11:51:02 where are the days of old unix systems where you'd just look for the best matching exploit to an outdated program :) 2014-04-03 11:52:29 I am afraid that with the distribution Alpine Linux is hard to find an exploit. 2014-04-03 11:52:36 yeah lol 2014-04-03 11:52:47 so we make our own DOSses 2014-04-03 11:52:56 :) 2014-04-03 12:21:18 lol 2014-04-03 14:30:28 c ya 2014-04-03 21:10:42 has anyone got ipv6 working with LXC ? 2014-04-03 23:07:22 BitL0G1c: inside an lxc guest? 2014-04-03 23:08:45 yes - the host has ipv6 connectivity - but the guest doesn't seem to 2014-04-03 23:09:07 given the guest a global ip 2014-04-03 23:09:24 BitL0G1c: most of my lxc guests are macvlan/bridged 2014-04-03 23:09:34 but i think they picked up a link local addr 2014-04-03 23:09:51 yes it has a link local too 2014-04-03 23:10:03 interface is a veth 2014-04-03 23:10:23 i can ping ip4 out - but not ping6 2014-04-03 23:10:39 on my current guests I don't have real ip6 connectivity 2014-04-03 23:11:21 the vlans I have proper ip6 connectivity on aren't the ones where I run lxc :/ 2014-04-03 23:11:23 yet 2014-04-03 23:13:38 will keep fiddling around 2014-04-03 23:24:08 I have openvswitch running as the bridge for LXC - works very well 2014-04-05 01:17:32 any chance of getting an option put into the installer for encrypted filesystems? 2014-04-05 02:09:40 Mp5shooter: drop out of guided at disk partition, create your luks container and mount it, finish install and edit grub config if boot or root is encrypted 2014-04-07 14:11:20 did anyone yet use the zfs bits in something production-like? 2014-04-07 14:23:18 i only tried it in testing env, and the io was horrible. 2014-04-07 14:24:08 but i guess that was probably a hw vs sw issue. 2014-04-07 14:24:37 thats a good thing to know anyway 2014-04-07 14:25:28 its getting more popular on linux, so i guess they are doing something right. 2014-04-07 14:27:56 i used it like 2 years ago, it was pretty unstable back then (and people told me how stable it was b/c they just don't even know what stable means) 2014-04-07 14:28:29 i think i'll do another test ride since the online compression and such would be pretty nice to have 2014-04-07 14:28:56 future wise its fs heaven. 2014-04-07 14:29:48 not sure if any alpine user is using it atm. 2014-04-07 14:29:53 i remember somebody talking about it. 2014-04-07 14:30:47 i'm trying to decide. if i put it to a serious test i'll write something about it 2014-04-07 14:31:08 i find something wrong with any option i put on the table. makes me crash :) 2014-04-07 14:31:14 make sure you use the exact same drives, this is what i remember what is important. 2014-04-07 14:31:32 oh 2014-04-07 14:31:35 good to know 2014-04-07 14:32:23 i used different drives, and my io was poor. probalby because of that. 2014-04-07 14:32:44 well zfs users also seem to have a mode like this 2014-04-07 14:32:52 rule #1: zfs is awesome 2014-04-07 14:33:00 rule 2: zfs is fast 2014-04-07 14:33:14 role 3: if any of the above doesn't apply, blame some hardware 2014-04-07 14:33:55 it seems without a battery backed sas ram disk you can never use zil without high performance impact 2014-04-07 14:34:13 and if if that doesn't help, you need more ram 2014-04-07 14:34:21 and then it's the controller :) 2014-04-07 14:44:46 there are some cheap slog sata devices available that work perfectly 2014-04-07 14:45:00 cheap might be a weird way of saying it but yeah 2014-04-07 14:46:59 http://eng.joyent.com/manufacturing/bom.html 2014-04-07 14:47:00 thats joyents manufacturing list for their servers, which are field proven setups 2014-04-07 14:54:29 they have their own slog device 2014-04-07 14:54:39 i think that will have some black magic smoke inside 2014-04-07 14:54:55 and two stec ssds 2014-04-07 14:55:08 so they have highend ssds plus probably some ramdisk thing for the slog 2014-04-07 14:58:13 rdenis: can you give a commonly available example? maybe that would be the thing i should add 2014-04-07 14:58:28 for example joyent uses 2014-04-07 14:58:28 50GB 2.5 SATA Mach16 SLC SSD 2014-04-07 14:58:35 which runs about 400ish$ 2014-04-07 14:58:40 grabbing a used zeusram is something i read about a few times 2014-04-07 17:12:53 It's it possible that the drivers from bootcamp are causing the blank screen? so far everytime I update the drivers it blanks the screen up 2014-04-07 17:13:06 if I keep it with the ones from windows, it works fine 2014-04-07 20:01:08 http://heartbleed.com/ :( 2014-04-07 20:50:45 patch for OpenSSL to bump to v1.0.1g sent to fix heartbleed exploit 2014-04-08 18:26:39 Julian Assange - Debian is owned by the NSA (about 20 minutes into his talk) https://www.youtube.com/watch?v=UFFTYRWB0Tk 2014-04-08 18:34:07 :o 2014-04-08 18:34:39 openvz needs an alpine template 2014-04-08 18:35:16 hm 2014-04-08 18:35:19 is there a way to downgrade a package 2014-04-08 18:38:52 to downgrade a pkg from testing / edge to main - change your repo pinning & 'apk fix pkgname' 2014-04-08 18:42:18 ah 2014-04-08 18:42:20 ok 2014-04-08 18:56:17 i.e edit /etc/apk/world & remove @edge from pkgname@edge 2014-04-08 18:56:49 & apk update && apk fix pkgname 2014-04-08 18:57:21 I was going to see if I was able to downgrade to openssl 1.0.1f to see what my web server exposed 2014-04-08 18:57:27 but the package is gone haha 2014-04-08 18:58:07 debian have still not updated 2014-04-09 13:14:14 edge is now musl libc 2014-04-09 14:16:21 I'm using LXC 1.0 on edge - so I should change my repos from main to track edge & stop pinning packages ? (or not upgrade anything until Main is also musl) 2014-04-09 14:18:17 edge/main is musl now 2014-04-09 14:18:20 h, 2014-04-09 14:18:21 hm 2014-04-09 14:18:33 so you used pinned edge on v2.7? 2014-04-09 14:18:39 we have 2 options there 2014-04-09 14:18:43 3 2014-04-09 14:19:07 1) you upgrade evertying to edge/musl (scary...) 2014-04-09 14:19:18 2) downgrade everything to v2.7 2014-04-09 14:19:37 (you'll downgrade lxc) 2014-04-09 14:19:49 3) we backport lxc-1.0 to v2.7/backports 2014-04-09 14:20:09 and you replace your pinned edge with a pinned backports 2014-04-09 14:20:30 you can do option 1 and 2 without anything more 2014-04-09 14:20:55 if you for some reason depend and need lxc 1.0, then we could add it to v2.7/backports 2014-04-09 14:21:51 I'll try downgrading to main first of all - then upgrade lxc 2014-04-09 14:23:13 ok 2014-04-09 14:23:25 you will need apk-tools-2.4.2 2014-04-09 14:23:41 remove your pinned repo and 'apk upgrade -U -a' 2014-04-09 14:23:45 should do it 2014-04-09 14:23:59 ok ;-) 2014-04-09 20:44:49 njargls as i get around to my own boxes i found the 2.6 systems with the auth bug are of course stuck on 1.0.1f ssl 2014-04-09 20:45:06 bug-cas-cade 2014-04-09 20:55:33 i could still run an update from config management even with the fried root pw 2014-04-09 20:55:59 so, i guess i should get in via key 2014-04-09 22:03:25 hello everyone, i'm trying to have the xen-pciback module load instead of the mpt2sas on boot in alpine 2.7 . the instructions on the web website don't exactly apply to alpine and i'm struggling a bit… Would someone know how to do this ? I'm new to Alpine. 2014-04-09 22:11:53 @pp-dom0-xen - put the kernel options in /etc/update-extlinux.conf & run 'apk fix linux-grsec' to rebuild the initramfs 2014-04-09 22:22:00 maybe adding pciback to modules= so it gets loaded before anything else 2014-04-09 22:22:25 @BitL0G1c : thanks for the help. apk fix fails as it cannot find my exact version on the repo. 2014-04-09 22:23:20 i have it in /etc/modules but it loads after mpt2sas has claimed the device 2014-04-09 22:24:24 do apk update && apk upgrade to sync the kernel with the repo 2014-04-09 22:24:58 3.10.33-1 is the current in main 2014-04-09 22:25:52 that's what i have but not the repo 2014-04-09 22:26:18 # uname -a Linux alpine-xen 3.10.33-1-grsec 2014-04-09 22:27:22 repo has linux-grsec-3.10.36-r0 in main 2014-04-09 22:28:16 did the upgrade to see if the new initd works 2014-04-09 22:28:32 actually the current is 3.10.36-0 (I've not rebooted yet) 2014-04-09 22:29:05 I just ran apk fix ok 2014-04-09 22:29:16 ok the upgrade took care of the mkinitfs for me 2014-04-09 22:30:32 shoot, looks like it killed my xen setup ?! 2014-04-09 22:32:23 did you reboot into the 3.10.36-0 kernel ? 2014-04-09 22:32:57 yep and can' connect to hypervisor or find xen modules anymore apparently 2014-04-09 22:34:52 alpine edge / main has changed to MUSL too (from uclibc) - I also did today 2014-04-09 22:35:01 apk add apk-tools 2014-04-09 22:35:18 'apk upgrade -U -a' 2014-04-09 22:35:41 (after first removing any pinned packages in /etc/apk/world) 2014-04-09 22:36:15 then after the upgrade repinning any packages I track from edge 2014-04-09 22:36:30 & apk upgrade 2014-04-09 22:37:14 this 2nd upgrade installed MUSL 2014-04-09 22:38:23 when you say removing pinned packages you mean having an empty /etc/apk/world ? 2014-04-09 22:38:47 no - I track LXC in edge 2014-04-09 22:39:08 so in /etc/apk/world I have "lxc@edge" 2014-04-09 22:39:35 ok, i just have a rather vailla system with xen on it 2014-04-09 22:39:38 so I ust removed the @edge 2014-04-09 22:39:40 vanilla 2014-04-09 22:42:48 see if you can see 'musl' at the end of /etc/apk/world 2014-04-09 22:43:53 nope, but rebooting with the right kernel i have less problems with xen already… let me check the modules 2014-04-09 22:46:09 in update-extlinux.conf i have "modules=sd-mod,usb-storage,ext4,xen-pciback" 2014-04-09 22:47:02 is where I thought it should go (& not in /etc/modules) 2014-04-09 22:47:27 and still running app fix shows i'm still running uclibc 2014-04-09 22:48:48 'apk add apk-tools' 2014-04-09 22:48:48 'apk upgrade -U -a' 2014-04-09 22:48:48 'apk upgrade' 2014-04-09 22:49:09 the 2nd upgrade installed musl for me 2014-04-09 22:50:02 redone it, no new install 2014-04-09 23:05:05 i was thinking of putting the module in the features.d of mkinitfs ? 2014-04-09 23:57:23 ok i'll call it a night and come back some other time. Thx @bitlogic 2014-04-10 00:04:04 np 2014-04-10 09:03:02 pah, updating 109 packages took almost 1.5 minutes now! I need to tune more! 2014-04-10 09:09:20 :D 2014-04-10 09:19:01 hm, i could have used "time ..." for my musl upgrade... but did not think about this before... :-/ 2014-04-10 09:19:32 but anyway, replacing more than 400 apk's was really fast too :-) 2014-04-10 09:22:07 it took me 1 hour 40 mins for 1022 packages 2014-04-10 09:22:14 but i think something is wrong with my network here... 2014-04-10 09:22:46 reinstall all 1000 pkgs took ~5mins last time iirc 2014-04-10 09:23:22 Receiving objects: 100% (613/613), 2.15 MiB | 150.00 KiB/s, done. 2014-04-10 09:23:31 150KiB/s 2014-04-11 00:43:04 is anyone running musl+edge and testing LXC? 2014-04-11 00:44:15 I tried it, and found that the container can't do its initialization and boot sequence properly. Things fail to bind mount from the host, so I don't get a /tmp/.X11 for throwing back X displays 2014-04-11 00:44:32 curious if anyone has any thoughts on that 2014-04-11 01:09:33 duck1: my latest stuff with LXC has been on alpine 2.7.5 - haven't tried edge, sorry 2014-04-11 01:16:58 jbilyk_: I've done edge-uclibc, and that works 2014-04-11 01:17:08 its edge-musl that's no-go for my LXC 2014-04-11 01:23:30 edge-musl also seems to break gpgme - fwknop-server now gives "this GPGME implementation does not support OpenPGP" 2014-04-11 03:09:41 BitL0G1c: I found that Pidgin's SSL library can't connect to Google anymore, so maybe there's a crypto bug? 2014-04-11 04:53:05 duck1, I have edge-musl doing LXCs on arm 2014-04-11 04:53:34 BitL0G1c, yes, gpgme seems broke. claws' gpg plugin is defunct too. 2014-04-11 04:54:01 duck1, that might openssl bug that got uncovered when we turned openssl freelists off 2014-04-11 04:54:08 i can commit a fix for it in a moment 2014-04-11 05:12:41 duck1, can you retry ssl tests (in few minutes after it's rebuilt) with the openssl i just pushed. it should fix the issue. 2014-04-11 06:14:20 BitL0G1c, re: gpgme breakage, which architecture? x86, x86_64 or armhf? 2014-04-11 06:21:23 hmm. seems gpgme recompile on x86 fixed things 2014-04-11 06:23:07 hmm 2014-04-11 06:23:15 i think we need to recompile edge/x86 2014-04-11 06:23:23 edge/x86_64 was recompiled earlier 2014-04-11 06:23:28 ncopa, ^^^ 2014-04-11 06:42:20 gpgme works on x86_64 2014-04-11 06:42:28 i think we might need rebuild world for x86 2014-04-11 06:42:53 yes. rebuild fixes it for x86 2014-04-11 06:48:31 i wonder why it happens 2014-04-11 06:48:46 it was probably built with some broke version of musl 2014-04-11 06:49:05 maybe we should look the datesamp when it was built 2014-04-11 06:49:16 and purge everything older that that date 2014-04-11 06:49:18 or so 2014-04-11 06:49:26 so force a rebuild 2014-04-11 06:51:09 fabled: i will clean up the build-edge-x86 a bit 2014-04-11 06:51:13 it has perl and others installed 2014-04-11 06:52:21 ok 2014-04-11 06:52:23 good 2014-04-11 06:53:49 179325 Dec 24 09:06 gpgme-1.4.3-r0.apk 2014-04-11 06:53:56 was build in christmas :) 2014-04-11 06:54:21 probably with broken gcc / out-dated musl headers 2014-04-11 06:54:48 ok. please rebuild world on musl x86 2014-04-11 06:54:56 at least everything older than one month 2014-04-11 07:01:52 crow: found the issue with oscam? 2014-04-11 07:39:11 clandmeter not realy,same version running, but no crash for a while, and i am still running inside the gdb 2014-04-11 07:39:30 clandmeter was there any update from oscam in edge? 2014-04-11 07:40:36 crow: i didnt push anything 2014-04-11 07:40:45 anything usefull commited? 2014-04-11 07:41:23 there were some webif improvments and such, but i didnt check it completly. 2014-04-11 07:41:50 i am thingking to update my alpine tu edge-musl, but i think there are still not musl iso? 2014-04-11 07:43:05 its coming 2014-04-11 07:43:11 ncopa will spin one soon 2014-04-11 07:45:14 crow: let me push latest changes to edge, and ill setup a lxc with oscam to test it. 2014-04-11 07:48:01 clandmeter then lets wait on ncopa for iso :) 2014-04-11 07:48:26 lxc iso ist with musl? or you have arm iso build by yourself? 2014-04-11 07:49:28 lxc does not need any iso 2014-04-11 07:49:31 i dont get it 2014-04-11 07:49:41 i run musl edge in a lxc container 2014-04-11 07:49:56 any edge = musl nowdays ;) 2014-04-11 07:52:30 ACTION hides has old uclibc edge box. 2014-04-11 07:54:23 crow: im going to enable ipv6 support again. I disabled it before because of vserver issues. 2014-04-11 08:12:22 crow: i push new oscam, but x86 is rebuilding so that will still take some time to finish 2014-04-11 10:19:28 clandmeter thank you :) 2014-04-11 10:19:41 clandmeter i cant use ipv6 at all but does not matter 2014-04-11 10:20:11 aslong it doesnt break anything its fine. 2014-04-11 10:20:24 in vserver it refused to start bcause of it. 2014-04-11 10:21:37 or i dont know what lxc is, that was reason i asked such question 2014-04-11 10:23:13 linux containers 2014-04-11 10:23:35 similar like vserver (if you know what that is) 2014-04-11 10:23:48 figured it out when you wrote that you run musl edge in a lxc container 2014-04-11 10:24:02 :) 2014-04-11 10:24:15 its fairly easy to setup 2014-04-11 10:24:21 and its nice to run tests in 2014-04-11 10:25:36 user mode linux v2 :) 2014-04-11 10:36:22 x86 openvswitch on MUSL also pulls in the build dependencies (pkg-config gcc etc) when installed - this did happen previously on uclibc on the latest git - but not on the current ovs 2.1.0 2014-04-11 11:16:56 BitL0G1c: what version of openvswitch do you get? 2014-04-11 11:17:07 2.1.0 2014-04-11 11:17:14 and it pulls in gcc pk-config? 2014-04-11 11:17:22 yes on musl 2014-04-11 11:17:29 apk info -R openvswitch 2014-04-11 11:18:17 try apk del pkg-config 2014-04-11 11:18:17 hold on 2014-04-11 11:18:24 and see what hold it back 2014-04-11 11:20:19 ncopa: x86/edge broken? 2014-04-11 11:20:34 depends: 2014-04-11 11:20:40 busybox 2014-04-11 11:20:49 libatomic.so 2014-04-11 11:20:57 libc-musl 2014-04-11 11:20:57 yeah 2014-04-11 11:21:09 libcrypto 2014-04-11 11:21:11 and i am rebuilding lots of stuff there now 2014-04-11 11:21:21 libgcc_s 2014-04-11 11:21:25 libssl 2014-04-11 11:21:40 ncopa: ok, ill wait. 2014-04-11 11:22:07 pkg-config deletes ok 2014-04-11 11:22:26 will try with the other unneeded stuff 2014-04-11 11:25:31 (binutils-libs binutils libgomp pkgconf pkgconfig gmp5 mpfr3 mpcc1 gcc) - this is the extra stuff it pulls in 2014-04-11 11:28:44 makes no sense 2014-04-11 11:28:49 why would it 2014-04-11 11:41:16 Pre-order your .NINJA domain today... 2014-04-11 11:41:52 alpinelinux.ninja 2014-04-11 11:43:04 email from eurodns: You're a force of nature with control over the elements, a grandmaster at gaming, a champion of cooking, a warrior of words, a master of martial arts. Basically, you're an expert in your field possessing special skills and wearing a mask. Here's some ideas to inspire you: heavymetal.ninja, knitting.ninja, slicingdicing.ninja, pingding.ninja, ghosthunting.ninja, sheepshearing.ninja, koffee.ninja. 2014-04-11 11:43:16 oh my :) 2014-04-11 11:43:31 whats next? 2014-04-11 11:44:46 there is also a .guru 2014-04-11 11:45:18 what about .addict? 2014-04-11 11:45:33 lol 2014-04-11 11:46:04 what money can buy.... 2014-04-11 12:02:41 .dicks 2014-04-11 12:21:02 .lol 2014-04-11 12:42:29 i'd like .not 2014-04-11 12:42:33 im-smart.not 2014-04-11 12:52:58 .did 2014-04-11 12:53:14 website-make-you.did 2014-04-11 12:54:46 .stupid 2014-04-11 12:55:12 i'd love to own .stupid. 2014-04-11 12:55:20 so much that belongs in there :) 2014-04-11 13:43:26 fabled: I'll get to it tonight, the openssl rebuild testing for edge-musl 2014-04-11 23:45:10 ncopa: what DE do you use on Alpine? I'm using XFCE4, but I'm wanting something a little lighter weight. 2014-04-12 00:01:35 pidgin is fixed since the update fabled made 2014-04-12 02:15:31 so /etc/TZ doesn't seem to be relevant for the musl builds passing timezone information to applications. Aside from setting a TZ environment variable, is there another way to get musl to know the timezone/ 2014-04-12 06:02:07 Hey, Never heard of Apline Linux before. Can someone post a link the Distro site? 2014-04-12 06:28:49 So, hm. Is there something going on with the edge repo? 2014-04-12 10:16:09 Obnoxious: besides the switch to musl? 2014-04-12 10:16:39 not running edge anywhere right now, so i don't know what you noticed :) 2014-04-12 13:16:16 darkfaded: there's some weird stuff with edge+musl where switching breaks some things. I've switched back and forth 3-4 times now, and it seems that some package dependencies have changed and a couple packages don't install the first shot. I run the apk.static upgrade -U -a a couple times and it usually gets it all fixed, but I can see folks having issues with the transition. 2014-04-12 13:29:34 you're not supposed to switch back and forth, it is only one direction that is supported 2014-04-12 14:21:19 well if the one direction don't work? ;) 2014-04-12 14:23:16 <_lb_> Hi guys! I have a strong change proposal to be done to the Alpine booting, with some rationale... 2014-04-12 14:23:32 <_lb_> You see.. I have many machines runing alpine that I can’t access physically... 2014-04-12 14:23:53 kaniini: I'm with darkfaded, I upgraded to edge+musl and things didn't work, so I went to edge-uclibc, and if that doesn't work I should be able to gracefully downgrade to 2.7 2014-04-12 14:24:20 <_lb_> And the fact that the Alpine console is frame-buffered by default makes it impossible to use IPMI’s Serial Over Lan to troubleshoot machines with networking problems... 2014-04-12 14:24:56 <_lb_> Is there a voluntary decission to use framebuffer on boot? I tried to disable it but boot parameters are ignored 2014-04-12 14:25:35 <_lb_> It is useful when you have a KVM, but as IPMI has become better and better we gave up KVMs some time ago 2014-04-12 14:25:54 <_lb_> Or a better question…. Has any of you disabled the framebuffer on boot? 2014-04-12 14:28:12 _lb_: isn't that a kernel thing and nothing to do with the startup? I thought the framebuffer with how the kernel is compiled and what modules you load, not an init thing. Of course maybe with KMS things changed 2014-04-12 14:32:57 <_lb_> it is a kernel parameter… vga=normal 2014-04-12 14:33:18 <_lb_> but the fact that it boots with FB means the kernel was compiled to boot as such 2014-04-12 14:33:25 <_lb_> (There is an option to do so) 2014-04-12 14:33:39 <_lb_> This is very impractical for servers that are remotely managed... 2014-04-12 14:33:55 give me a minute to look at my setup 2014-04-12 14:34:00 <_lb_> The think is that I’ve tried to modify syslinux’s boot params 2014-04-12 14:34:06 <_lb_> And it ignores my flag 2014-04-12 14:34:48 duck1: that is what you get for running edge; if you are unwilling to accept lack of QA, then you shouldnt run edge 2014-04-12 14:35:39 _lb_: you can create an initramfs with kms disabled already 2014-04-12 14:35:45 _lb_: see /etc/mkinitfs.conf 2014-04-12 14:35:59 kaniini: oh, I'm perfectly fine running on edge, but the ability to switch between releases would be nice, i.e. have packages pinned to repo+version rather than just version. 2014-04-12 14:36:03 <_lb_> Sure, but my proposal is not to make it a default... 2014-04-12 14:36:12 <_lb_> Perhaps I should post this to the list... 2014-04-12 14:36:16 _lb_: not gonna happen 2014-04-12 14:36:29 _lb_: you're running impi is a server deployment, that's rather specialized 2014-04-12 14:36:38 <_lb_> What? 2014-04-12 14:36:41 <_lb_> Specialized? 2014-04-12 14:36:47 ipmi 2014-04-12 14:36:59 and, it's not impossible to use ipmi sol 2014-04-12 14:37:09 you just have to properly configure it instead of depending on bios redirection 2014-04-12 14:37:09 I've never seen a desktop or laptop computer with ipmi or a roamer card or anything similar, only servers 2014-04-12 14:37:14 <_lb_> kaniini: there is another alternative.. 2014-04-12 14:37:29 <_lb_> It would be to modify inittab so it has a second serial console attached 2014-04-12 14:37:45 <_lb_> I thought Alpine was focused on networking... 2014-04-12 14:37:51 _lb_: unnecessary, use console= 2014-04-12 14:38:48 <_lb_> I don’t mean to troll or anything... 2014-04-12 14:38:57 <_lb_> But please let me know if I am mistaken 2014-04-12 14:39:48 our use of kms has nothing to do with your inability to properly configure a commandline 2014-04-12 14:40:05 console=ttyS1,19200n8r 2014-04-12 14:40:07 very simple 2014-04-12 14:40:26 <_lb_> ACTION actually has not tried that one... 2014-04-12 14:41:05 <_lb_> In Solaris I have to do something similar to get double redirection… Let me see if it flies here 2014-04-12 14:41:10 well, it is the correct way to do it 2014-04-12 14:41:40 <_lb_> IMHO it is a phylisophical discussion we are not going to have, you have made that clear 2014-04-12 14:42:06 <_lb_> s/phylisophical/phylosophical/g 2014-04-12 14:42:25 <_lb_> But it is a very appropiate fix. Let me test it. 2014-04-12 14:44:18 duck1: you can do that already with repository tags 2014-04-12 14:44:28 duck1: just add the legacy uclibc edge with a tag 2014-04-12 14:44:37 duck1: then, apk add blah@uclibc 2014-04-12 14:44:39 or whatever 2014-04-12 14:44:44 you just put @foo infront of the url 2014-04-12 14:52:15 kaniini: sweet, thanks 2014-04-12 14:52:28 <_lb_> ACTION pxebooting now 2014-04-12 14:52:51 <_lb_> and… it ignored me again! 2014-04-12 14:52:53 <_lb_> Weeee 2014-04-12 15:00:51 darkfaded: the repo for every mirror had an untrusted signature. 2014-04-12 15:01:07 not sure if it's still like that 2014-04-12 15:01:19 I gave up on trying to fix thing and reimaged. 2014-04-12 15:03:26 <_lb_> kaniini: I think I am gonna have to customize the image 2014-04-12 15:03:36 <_lb_> no way jose! 2014-04-12 15:11:19 <_lb_> on the other hand… silly question... 2014-04-12 15:11:49 <_lb_> Some other distros allow to set a password for root on boot… (hashed or in plain text, gulp!) 2014-04-12 15:11:57 <_lb_> I know this machine is getting ssh on boot 2014-04-12 15:12:16 <_lb_> but I can’t login because ssh won’t let me unless root has a password defined... 2014-04-12 15:12:27 <_lb_> Is there a flag to pass such a string 2014-04-12 15:14:11 <_lb_> or…. Mmmm… 2014-04-12 15:22:33 apkovl can work on cdrom 2014-04-12 15:22:54 <_lb_> kaniini: that’s what i was thinging! 2014-04-12 15:22:58 <_lb_> Thanks man! 2014-04-12 15:23:15 <_lb_> s/thinkging/thinking/g 2014-04-12 15:23:25 <_lb_> apologies, I only have one hand… 2014-04-12 15:23:33 <_lb_> ACTION makes a lot of typos 2014-04-12 16:13:25 which pkg contains /sbin/ldconfig ? (I'm reverting a musl-edge lxc guest to uclibc-edge) 2014-04-12 16:15:07 I tried using an LXC guest with musl-edge as a build environment - but configure could not find a c compiler 2014-04-12 16:16:16 (gcc & g++ were both installed) 2014-04-12 16:35:29 going to create a new guest - reinstalling alpine-base didn't fix it 2014-04-12 17:44:40 what's the pastebin app that its recommended to use? 2014-04-12 17:55:56 sprunge ? 2014-04-12 18:08:53 command line pastebin - https://github.com/rupa/sprunge 2014-04-12 18:09:25 http://alpinelinux.org/apk/main/x86/sprunge 2014-04-12 18:25:47 hi, any ideas why xens pciback.hide won't work for me? 2014-04-12 18:26:00 i put xen-pciback in /etc/modules 2014-04-12 18:26:06 and updated the initrd 2014-04-12 18:26:24 and then i put the device to steal in a file in /etc/modprobe.d 2014-04-12 18:26:43 and i now also tried to blacklist the "other" driver, namely megaraid_sas 2014-04-12 18:27:04 do i need to do anything after i add something to /etc/modprobe.d/blacklist? 2014-04-12 18:27:08 like, new initrd or idk 2014-04-12 18:27:21 daveh0003:~# cat /etc/modprobe.d/xen-pciback.conf 2014-04-12 18:27:21 options xen-pciback hide=(0000:01:00.0) 2014-04-12 18:27:35 anything obviously wrong with that? 2014-04-12 18:27:50 daveh0003:~# tail -2 /etc/modprobe.d/blacklist.conf 2014-04-12 18:27:50 blacklist megaraid_sas 2014-04-12 18:32:57 grmbl: now i have this in the kernel command line: 2014-04-12 18:32:58 modules=sd-mod,usb-storage,ext4 modprobe.blacklist=megarai 2014-04-12 18:32:59 d_sas 2014-04-12 18:33:27 and it's still loading that darn driver 2014-04-12 22:04:27 BitL0G1c: I thought there was a package alpine had that was recommend for pasting stuff to rather than spamming the IRC channel, sprunge may be it, but I thought it was something else 2014-04-12 22:13:58 sprunge is good 2014-04-12 22:16:55 Can anyone assist me with debugging the time settings on Alpine+musl that I'm seeing here http://sprunge.us/IDVD? 2014-04-13 14:07:24 kaniini & ncopa: I'm successfully running edge-musl. I pinned edge-uclibc and have lxc@edge-uclibc lxc-libs@edge-uclibc as the only things I need pinned to the old uClibc to work properly. Thanks for your hard work on getting everything synced up and working. 2014-04-13 17:57:35 clandmeter today i have hat again segfault, and i was not able to catch crash with gdb :( 2014-04-13 18:04:31 clandmeter and new oscam is borked 2014-04-13 18:09:40 clandmeter ncopa here at the end of paste is new oscam build error when starting. and on top is tryout to catch segfault but gdb cant do it. http://pastebin.com/BZu80q3t is there any hope that i catch this segfault with gdb, before i was able (not used alpinelinux in that time) but now there is something wrong.. 2014-04-13 18:22:20 clandmeter i was trying to get svn version instead r0 but i did not sucess, why is that problem? 2014-04-13 18:40:49 isnt oscam for pirating dvb service 2014-04-13 18:46:35 kaniini: might be also for cctv recording, i think i heard it in that context too 2014-04-13 18:46:58 i... i'm not so sure on that 2014-04-13 18:47:02 hehe 2014-04-13 18:47:06 http://www.streamboard.tv/oscam/ 2014-04-13 18:47:20 this seems like the damnedest cctv recording software i've ever seen 2014-04-13 18:47:42 oh lol ok that seems pretty single tasked 2014-04-13 18:48:00 i don't like watching tv very much 2014-04-13 18:49:23 kaniini: btw i had a lot of fun reading your posts on WHT about the dtc panel 2014-04-13 18:51:11 darkfaded: i hate thomas 2014-04-13 18:54:13 hehe 2014-04-13 18:54:27 he surely knows how to get in his own way 2014-04-13 18:57:09 oh oh oh 2014-04-13 18:57:21 darkfaded: link por favor 2014-04-13 18:59:07 Obnoxious: for the thread? 2014-04-13 18:59:24 yeah 2014-04-13 18:59:28 sec 2014-04-13 19:00:39 iirc this one 2014-04-13 19:00:39 http://www.webhostingtalk.com/showthread.php?p=9019584#poststop 2014-04-13 19:01:18 there's one post where someone hands him a list of 5 exploits he searched just because of being told he'd not find any 2014-04-13 19:01:22 "but this is secured by" 2014-04-13 19:01:24 "no" 2014-04-13 19:01:26 "but then" 2014-04-13 19:01:28 "no" 2014-04-13 19:02:11 the best is he always says "no guarantees of security without sbox mode" - that mode is only there for like 2 years and so hella broken to setup... 2014-04-13 19:02:35 so he's basically saying any time from 2003 till 2013 it was completely insecure :) 2014-04-13 19:03:14 and of course it ends not with "ok i'll fix those 5 asap!" but ... silence 2014-04-13 19:09:46 lmao 2014-04-13 20:44:05 kaniini you can call it that, i will call a tool which let me use my subscribtion card which i bought. 2014-04-13 20:51:14 the problem wich i have is segfault and way to get gdb catch this segfault. could be also some other application. 2014-04-13 21:37:06 crow: whats wrong with new oscam? 2014-04-13 22:49:13 darkfaded: well 2014-04-13 22:49:26 darkfaded: i used to work for the guy who really did most of the ripping apart of dtc 2014-04-13 22:49:36 darkfaded: we always kind of wanted to, but never had any particular reason to do so 2014-04-13 22:49:56 darkfaded: i mean, nobody uses that crap so why would anyone care 2014-04-13 22:59:46 alpine-sdk for x86 MUSL appears to be missing. Openvswitch in the repos on MUSL is still pulling in gcc + build dependencies. When I build it in a MUSL LXC build environment it does not pull them in. When I installed the alpine-sdk dependencies - I did notice MUSL does not pull in libgcc (uclibc does) - perhaps this has something to do with how abuild sets the dependencies ? 2014-04-14 07:04:15 clandmeter i can not start it here anymore, i am on stable + oscam from edge. but one before latest was running just fine, you can see end of that paste which i posted above 2014-04-14 07:05:38 crow: do you know we have moved the musl? 2014-04-14 07:05:51 you cannot mix edge repo with 2.7 2014-04-14 07:06:03 you will have to rename it to edge-uclibc 2014-04-14 07:06:16 s/the/to 2014-04-14 07:06:43 weird sentence when i read that back :) 2014-04-14 07:19:38 crow: This paste has been removed! so i couldnt look at it. 2014-04-14 07:30:10 clandmeter i know that you moved to musl, and now i understand also why it does not work :) 2014-04-14 07:30:19 it was long day yesterday :D 2014-04-14 07:30:32 :) 2014-04-14 07:31:00 clandmeter hmm i was sure i put 1 day to stay alive. 2014-04-14 07:31:01 pinned repo's are tricky 2014-04-14 07:31:30 for sure when abi changes 2014-04-14 07:34:11 clandmeter i fully understand :), cant wait for edge-musl iso :) 2014-04-14 07:34:35 clandmeter segfault and way how to catch it is what is for me more interesed. look here http://pastebin.com/i1qrEHnV 2014-04-14 07:34:49 you will see segfault but in gdb it exited normaly 2014-04-14 07:35:03 so i think it have something to do with the threads 2014-04-14 07:37:46 crow: you still use grsec kernel? 2014-04-14 07:38:22 it looks like pax killed gdb 2014-04-14 07:40:18 clandmeter i am still using grsec kernel as it is default. 2014-04-14 07:40:54 did you use paxutils on oscam-debug? 2014-04-14 07:41:06 yes 2014-04-14 07:41:47 im no pax expert by far. 2014-04-14 07:42:00 ncopa can probably say more about those msg's 2014-04-14 07:42:19 clandmeter, crow: do note that x86 and other arches might need different paxflags 2014-04-14 07:42:29 see e.g. firefox apkbuild for the use 2014-04-14 07:42:42 x86 usually needs more pax flags 2014-04-14 07:43:08 thx fabled for the headsup 2014-04-14 07:43:21 one line paste: 2014-04-14 07:43:22 paxctl -c -m /usr/bin/gdb 2014-04-14 07:43:22 file /usr/bin/gdb had a PT_GNU_STACK program header, converted 2014-04-14 07:43:30 ok two :) 2014-04-14 07:43:55 i used that fev weeks ago,and also on /root/bin/oscam.debug 2014-04-14 07:44:10 not sure its wise to run it in gdb when pax kills it. 2014-04-14 07:44:27 maybe its better to generate a core and then use gdb. 2014-04-14 07:44:44 i could be wrong (not much experiance with gdb) 2014-04-14 07:44:46 if it's musl -- yes, create core 2014-04-14 07:44:55 musl + gdb do not fully work yet together 2014-04-14 07:45:00 it's known limitation 2014-04-14 07:45:10 he is still on uclibc 2014-04-14 07:45:15 fabled and 2.7.5 ? 2014-04-14 07:45:32 fabled: but pax kills it. i guess its better not to run it in gdb and let it kill gdb? 2014-04-14 07:46:05 yeah, gdbing core works generally better 2014-04-14 07:46:41 let me check how you managed pax flags on firefox. 2014-04-14 07:47:40 crow: which paxctl flags did you use? 2014-04-14 07:47:54 clandmeter paxctl -c -m /usr/bin/gdb 2014-04-14 07:48:09 http://git.alpinelinux.org/cgit/aports/tree/main/firefox/APKBUILD#n112 2014-04-14 07:49:10 crow: use those flags on oscam.debug, and let it run 2014-04-14 07:49:37 and dont run it in gdb, and make sure you set ulimit to generate core. 2014-04-14 07:51:21 crow: and if you want to run latest oscam, you can run lxc on 2.7.5 and run it in a container. works pretty nice. (im doing it now for testing reasons). 2014-04-14 07:55:27 clandmeter can an ALIX 3D3 500MHz 256MB Ram, do something like lxc ? 2014-04-14 07:56:09 sure 2014-04-14 07:56:41 i dont think lxc brings a lot of overhead 2014-04-14 07:56:48 it just isolates processes 2014-04-14 07:56:56 in a smart way 2014-04-14 07:57:14 crow, no special cpu supported needed. lxc containers are kernel built-in feature 2014-04-14 07:57:19 cool :) 2014-04-14 07:57:25 so yes. as long as you have enough memory. it's ok 2014-04-14 07:57:29 fabled i was thinking on low power from that cpu 2014-04-14 07:57:41 yeah. i have alix boxes here 2014-04-14 07:57:45 not doing lxc though 2014-04-14 07:57:57 the new apu boxes look nice too 2014-04-14 07:58:45 clandmeter ok if i get it correctly in that APBUILD fabled used this: paxctl -c -msp oscam.debug 2014-04-14 07:59:02 no i dont think so 2014-04-14 07:59:18 [ "$CARCH" = "x86" ] && paxflags="-msp" 2014-04-14 07:59:25 -msp 2014-04-14 07:59:39 having -msp is ok for any arch 2014-04-14 07:59:45 but in most arches just -m is enough 2014-04-14 07:59:53 x86 being the exception and requiring all three flags 2014-04-14 08:00:25 if thats true, that would explain why he keeps having those crashes 2014-04-14 08:00:30 clandmeter yes and then paxctl -c $paxflags 2014-04-14 08:00:54 oh yes sorry 2014-04-14 08:00:59 i didnt read carfully 2014-04-14 08:01:24 the idea is to add minimal pax flags tomake things work 2014-04-14 08:03:14 fabled and clandmeter ok i did this new flags on oscam.debug 2014-04-14 08:03:39 wonder if i can do it in init script this ulimited 2014-04-14 08:04:06 why would you want it in init? 2014-04-14 08:06:30 well ulimit -c unlimited is active only in current shell. and if i reboot server i need to set it again right. 2014-04-14 08:10:11 yes 2014-04-14 08:10:23 i think you can find some good info on google about it 2014-04-14 08:10:35 set it on boot, or per process 2014-04-14 08:24:21 crow: are you using dvb-s? 2014-04-14 08:35:38 clandmeter searching for it, all howtos which i find are talking about this file /etc/security/limits.conf .. 2014-04-14 08:35:43 clandmeter yes dvb-s and dvb-s2 2014-04-14 08:38:41 crow: i just bought this box: http://www.gss.de/en/products/?no_cache=1&tx_sytgssproduktdatenbank_pi3%5Bcat%5D=794&tx_sytgssproduktdatenbank_pi2%5Bcat%5D=794&tx_sytgssproduktdatenbank_pi1%5Bcat%5D=794&cHash=52bc7fee4e531e7f9a27e329ca1204d0 2014-04-14 08:40:35 looks interesting. what is backend inside? 2014-04-14 08:40:50 i am using VDR on my Client box 2014-04-14 08:41:06 which is actualy server/client instnc 2014-04-14 08:41:28 not sure what is inside 2014-04-14 08:41:32 its tuner only 2014-04-14 08:41:53 im using tvheadend, and they are adding sat>ip support. 2014-04-14 08:42:22 so it acts like a regular tuner over ip. it should leave all the DVB stream in tact. 2014-04-14 08:43:02 which means you can still decrypted encrypted channels in your server. 2014-04-14 08:43:40 and 120 euro for 4 tuners, is not that expensive. 2014-04-14 08:44:22 and no more problems with dvb kernel driver madness 2014-04-14 08:46:03 yea that realy cheep, and OS is Linux so it must be something interesting. 2014-04-14 08:47:34 never used tvheadend, started using VDR 6 years before on linux pc, and never stoped it. 2014-04-14 08:51:33 clandmeter but i dont see any CI slots or something like that for cards? 2014-04-14 08:51:54 it doesnt have any 2014-04-14 08:52:06 its a tuner 2014-04-14 08:52:10 so we need to use oscam, and thus i find it legal :) 2014-04-14 08:52:24 there is nothing illegal about oscam 2014-04-14 08:52:41 answer on '(20:40:49) (@kaniini) isnt oscam for pirating dvb service' 2014-04-14 08:52:45 that would make windows media player also illegal :) 2014-04-14 08:53:00 it can copy cd's to your computer ;-) 2014-04-14 08:53:12 i find it also not, 2014-04-14 08:53:20 its how you use it what defines legal/illegal 2014-04-14 09:12:56 exactly :) 2014-04-14 09:19:01 so, http://wiki.alpinelinux.org/wiki/Edge mentions --available to upgrade all packages, regardless of their version numbers ... my 2.7.5 image's apk does not have any --available switch. every time I try and go edge, some executable is missing some library stuff. 2014-04-14 09:23:37 the question is: how to upgrade all packages, even if their version numbers match? -- the underlying question is: how to upgrade to edge? 2014-04-14 09:26:29 Aspi: fabled just updated apk-tools in 2.7 2014-04-14 09:27:07 added or removed --available? 2014-04-14 09:27:23 Aspi edge is now musl so maybe not everything can be updated http://alpinelinux.org/edge-musl 2014-04-14 09:28:49 Aspi: you want to update to edge from stable? 2014-04-14 09:28:55 I see. thanks 2014-04-14 09:29:08 in other words, I stumbled upon Alpine at the worst possible time? :) 2014-04-14 09:29:26 depends what you want to do. 2014-04-14 09:29:38 2.7 is our stable branch based on uclibc 2014-04-14 09:29:52 Aspi: i think it would be best to run stable for 2-3 more weeks even if there's something interesting in edge 2014-04-14 09:29:55 keep using that repo is conciderd safe 2014-04-14 09:30:23 and edge-uclbic can be used to mix edge and stable. 2014-04-14 09:30:29 what I want to do is to create a safe, frugal USB install. alpine looks very appropriate 2014-04-14 09:30:32 edge-uclibc 2014-04-14 09:30:36 I saw 2014-04-14 09:30:43 Aspi and there is one blog from http://kaniini.dereferenced.org/2014/03/23/upgrading-to-alpine-musl.html 2014-04-14 09:31:59 alpine runs nicely on usb. just copy the contents from cd to usb, clear the mbr and install syslinux. (thats how i do it) 2014-04-14 09:32:14 s/cd/iso 2014-04-14 09:32:35 hah 2014-04-14 09:33:06 yeah, Unetbootin also fixed that problem for me 2014-04-14 09:34:05 in particular, I'd like to run icedtea and openjdk-7-jre on it. I notice that is not in the stable repos as well. is there any estimates on when it will be considered stable anywhere? http://alpinelinux.org/apk/main/x86_64/icedtea-web-mozilla 2014-04-14 09:36:06 Aspi: not sure what the current status is. fabled did a lot of work on it to make it work on musl. 2014-04-14 09:37:08 not sure he make it work on uclibc. i remember there were some issues. 2014-04-14 09:37:21 fabled: ^ 2014-04-14 09:37:27 mkay. just found http://alpinelinux.org/apk/main/x86/openjdk7 as well, it mentions icedtea (without having it as a dependency), the relationship between the packages are? 2014-04-14 09:38:01 i think we will not bother backporting openjdk-7 to -stable 2014-04-14 09:38:02 yeah, on my stable 2.7, installing icedtea-web-mozilla (with openjdk-6-jre) crashes firefox 2014-04-14 09:38:07 2.7 that is 2014-04-14 09:38:17 it seems to work ok for what i tested it on edge 2014-04-14 09:38:44 icedtea is the build scripts, and set of patches that are used to build openjdk 2014-04-14 09:39:12 ^that are specific for web browsers? 2014-04-14 09:40:47 and is openjdk-7 due to next release, when musl and more have settled? 2014-04-14 09:44:23 Aspi, yes, openjdk-7 will be automatically part of next major release 2014-04-14 09:44:57 basically anything edge will become the next release when it's being branched 2014-04-14 09:44:58 major == 3.0? 2014-04-14 09:45:04 in this case yes 2014-04-14 09:45:37 but generally, in alpine-x.y.z; x.y is the 'major' version 2014-04-14 09:46:17 okat 2014-04-14 09:46:26 s/t/y 2014-04-14 09:46:32 (I'm learning, aye?) 2014-04-14 09:48:56 tested with edge-uclibc: icedtea-web-mozilla still crashes firefox 2014-04-14 09:49:08 hm 2014-04-14 09:49:12 it probably needs a rebuild 2014-04-14 09:49:30 in fact, ncopa, we should drop openjdk6 and rebuild anything requiring java against openjdk7 in current edge 2014-04-14 09:50:47 count one in favor for that idea :) 2014-04-14 09:52:07 if openjdk7 is built with/through icedtea, it should be usable by web browsers? 2014-04-14 09:57:06 icedtea-web-mozilla is from separate source package that provides the plugin 2014-04-14 10:05:55 well, guess there's not much more I can do than eagerly wait for 3.0. keep going, I love the ideas behind Alpine. 2014-04-14 10:06:26 i think 3.0 is planned for early May 2014-04-14 10:07:06 http://wiki.alpinelinux.org/wiki/Alpine_Linux:Releases 2014-04-14 10:07:17 it's 6mo release cycle, and previously it's been always early May 2014-04-14 10:07:38 though. i think we will declare 3.0 'beta' quality due to switch to musl, and many other major changes 2014-04-14 10:07:55 that should allow us to do the release in May 2014-04-14 10:08:04 nice. I'm looking forward to it 2014-04-14 10:08:38 i'm off for the after noon 2014-04-14 10:09:45 crow: http://git.alpinelinux.org/cgit/aports/commit/?id=aeae89ebfb6d01961dfad95de155e0b82a6ad75e 2014-04-14 10:14:00 clandmeter thats now in stable? or edge-ulibc? 2014-04-14 10:14:24 its in edge 2014-04-14 10:14:41 ok reading about lxe :D 2014-04-14 10:14:47 all developement is happening in edge 2014-04-14 10:14:55 lxc ;) 2014-04-14 10:15:05 erm lxc 2014-04-14 10:15:09 or you are on a differnt subject now :) 2014-04-14 10:15:22 nop, you told me to use lxc on stable :D 2014-04-14 10:15:48 in edge you can now install oscam-dbg which *will* contain symbols 2014-04-14 10:16:29 crow: our wiki has a simple howto about lxc. 2014-04-14 10:16:52 will remove pinned repos to 2014-04-14 10:17:29 clandmeter i saw diff and code about not to stipp stuff 2014-04-14 10:18:10 yes, oscam wants to be smart and strip by itself 2014-04-14 10:19:23 oscam is not designed for packagers (distro usage) and they probably dont care. 2014-04-14 10:23:48 clandmeter why here i have warrning from @edge oscam even i delete it before http://pastebin.com/P6hJ9bJ3 2014-04-14 10:25:08 crow: do not specify the repo when you delete 2014-04-14 10:25:27 ok 2014-04-14 10:25:35 you can only hold one pkg per pkgname in your apk db. 2014-04-14 10:26:03 apk del oscam oscam-dbg should be enough 2014-04-14 10:27:09 i wonder if its better to output a msg instead of nothing when pkg is not found to delete. 2014-04-14 10:28:19 without specify repo it works, thanks 2014-04-14 10:28:46 as i saw OK i was thingkingthat it is deleted 2014-04-14 10:28:57 i would vote for msg pkg not found to delete 2014-04-14 10:35:14 hmm my date is worng (actually time) i am using chronyd with default settings, and chronyd is started but it will not sync to correct time 2014-04-14 10:39:06 ^* portal2.theaterservice.at 3 6 37 12 +8599ns[ +333us] +/- 81ms 2014-04-14 10:39:13 chronyc > sources 2014-04-14 11:00:28 timezones are different in musl 2014-04-14 11:16:25 ncopa: did timezone corrections for musl already get merged? 2014-04-14 14:21:19 ncopa: ? 2014-04-14 14:24:36 clandmeter i am still on stable, 2014-04-14 14:25:13 clandmeter i am two hourse behinde.. Mon Apr 14 14:24:48 UTC 2014 (16:25:10) 2014-04-14 14:34:33 clandmeter i set system time to UTC and zone to Europe/Vienna, 2014-04-14 14:35:08 crow: its a know issue 2014-04-14 14:35:16 musl does timezones differently 2014-04-14 14:35:19 more simple 2014-04-14 14:35:42 i think you can copy your timezone file to etc/somefile 2014-04-14 14:36:37 http://wiki.musl-libc.org/wiki/Environment_Variables 2014-04-14 14:37:03 crow: install tzdata 2014-04-14 14:37:27 and link your location in /usr/share/timezone (i guess this is the path) to /etc/zoneinfo 2014-04-14 14:39:20 yea but i am still on glibc system 2014-04-14 14:39:39 oh 2014-04-14 14:39:51 setup-timezone should do it, and i isntalled chrony 2014-04-14 14:40:19 and time still off? 2014-04-14 14:42:04 well setup-timezone -z Europe/Vienna did it, but on laptop i have UTC but it sync the correct time 2014-04-14 14:42:15 now it is: Mon Apr 14 16:41:07 CEST 2014 2014-04-14 14:46:20 clandmeter on another box (archlinux) i set hardware clock to UTC; and then enabled NTP so it is actually not the same which i wrote here 2014-04-14 14:47:29 crow: your issue now is? 2014-04-14 14:50:02 non i set timezone to correct one and it works. 2014-04-14 14:50:11 ok good 2014-04-14 14:50:29 if you setup lxc with musl, then it wont work :) 2014-04-14 14:51:01 you need to copy zonefile to /etc/TZ and export TZ=/etc/TZ 2014-04-14 14:51:26 just for a heads up 2014-04-14 15:00:44 Ok thanks :) 2014-04-14 15:15:48 is there a documenation about howto install alpine onto a raspberry pi (armhf) ? 2014-04-14 15:44:05 mirac: i think there's just some tarball for a sd card 2014-04-14 15:50:25 darkfaded: apk resolves to the armhf packages? 2014-04-14 15:51:22 i havent tried it :( 2014-04-14 15:51:24 it should 2014-04-14 15:51:38 would make sense... 2014-04-14 15:51:39 well, same as for 32/64 bit 2014-04-14 15:52:03 where can i find the tar-ball ? 2014-04-14 15:53:31 couldn't find it right now, but i guess the actual dev will wake up at some time 2014-04-14 15:53:49 i only searched a minute or so, so if you try smarter or longer it should be somewhere 2014-04-14 15:54:11 i think in a month or so those will be regular builds 2014-04-14 15:58:42 ah... okey.. thanks 2014-04-14 15:59:23 looks good: http://dev.alpinelinux.org/~tteras 2014-04-14 16:02:38 ah nice 2014-04-14 16:02:51 there's a arm cpu thread on -devel 2014-04-14 16:02:56 you might wanna post there 2014-04-14 16:11:06 i will try it first... thank you darkfaded 2014-04-14 22:10:59 how do you guys rank LXC right now, is there any real benefit over a plain chroot (on a grsec box after all) from a security point of view? 2014-04-14 22:54:13 @darkfaded - LXC is much easier to maintain than a chroot & it can have it's own firewall. Version 1.0+ also runs as an unprivileged process 2014-04-14 22:54:36 it also makes a great alpine build environment 2014-04-15 00:44:06 I was an idiot and managed to remove everything in /dev. Where does Alpine populate /dev? 2014-04-15 01:05:49 @Xe - /dev - is a virtual filesystem - you just need to reboot the box, or if you use udev - restart udevd - both of these will repopulate /dev 2014-04-15 01:12:31 BitL0G1c: That's kinda not an option 2014-04-15 01:12:44 I don't know if the box will come back 2014-04-15 01:41:11 remount /dev? 2014-04-15 01:41:23 I have 2014-04-15 01:41:28 Didn't fix it 2014-04-15 01:41:39 Is there a way to validate an extlinux configuration? 2014-04-15 01:43:29 prolly should be within the extlinux pkg itself 2014-04-15 01:44:40 Reasons I love linux: You hear awesome stories of people handling shit like this 2014-04-15 01:44:52 like rm -rf / 2014-04-15 01:44:56 as root 2014-04-15 01:45:26 and recovering while the system kept running 2014-04-15 01:47:27 bah 2014-04-15 01:47:36 there's no obvious tool for it 2014-04-15 01:47:40 Xe: quick google reveals back in 2010 /dev was populated by busybox's mdev 2014-04-15 01:51:26 mdev -s 2014-04-15 01:54:47 oh yay 2014-04-15 01:55:15 did it work? 2014-04-15 04:29:27 Diftraku: yes 2014-04-15 14:32:42 http://hastebin.com/korehawahe.avrasm 2014-04-15 14:52:04 alacerda: are acf-openldap included in v2.7? 2014-04-15 19:37:14 so will the next release of Alpine be based fully on musl? 2014-04-15 19:38:18 3.0 will be musl based 2014-04-15 19:40:31 is 3.0 going to be the next release, or will 2.7 continue 2014-04-15 19:40:37 what is the time frame for 3.0? 2014-04-15 19:41:57 jackv: theoretically, 3.0 comes out in May 2014-04-15 19:44:06 ok, thank you! 2014-04-15 19:44:17 jackv: they will run parallel but eventually only sec updates will be pushed to 2.7 2014-04-15 19:45:03 2.7 has backports now. 2014-04-15 19:46:25 so 2.7 is the last 2.x release 2014-04-15 19:49:12 yes 2014-04-15 19:51:08 musl is a big change for alpine. some users will probalby prefer to keep 2.7 for a while. 2014-04-16 01:00:46 hi few questions: is alpine going to be an alternative to openwrt? 2014-04-16 01:01:36 and does it currently work on routers? 2014-04-16 06:23:15 nwmcsween: if you mean mips based consumer routers then no, I don't think there are any plans for that 2014-04-16 06:31:03 morning. correct. no plans for mips currently 2014-04-16 06:31:12 we are working on arm though 2014-04-17 08:57:39 Hi! 2014-04-17 08:58:00 Anybody knows if there's other VNC viewers than gvncviewer in AL? 2014-04-17 09:09:13 strange nobody has packed tightvnc 2014-04-17 09:12:06 could add tigervnc probably? 2014-04-17 09:14:51 clandmeter: yes, better than tightvnc (which has been unmaintained for some time AFAIK) 2014-04-17 09:20:51 15 patches in fedora for tigervnc... 2014-04-17 10:28:58 hello yes I am on edge and am receiving an error when I attempt to install ca-certificates from any mirror (tried nl, dl-2, dl-3) 2014-04-17 10:29:01 (1/1) Installing ca-certificates (20140325-r0) 2014-04-17 10:29:04 0% [ ]wget: short write 2014-04-17 10:29:06 ERROR: ca-certificates-20140325-r0: BAD signature 2014-04-17 10:29:16 it's kind of messing up curl rather royally 2014-04-17 10:29:27 this happens on two testing boxes and in a brand new freshly made chroot 2014-04-17 10:29:58 awilfox, 'apk update' first 2014-04-17 10:30:16 x86? 2014-04-17 10:30:20 fabled, as I said this happens in a freshly made chroot, there is nothing to update, it fetched the index seconds ago 2014-04-17 10:30:26 x86_64 2014-04-17 10:30:33 on all three machines 2014-04-17 10:30:48 sounds like the apkindex and the .apk is out-of-sync 2014-04-17 10:31:53 yes, because downloading the apk manually from the mirror and `apk add` on the raw file succeeds 2014-04-17 10:32:14 just really bizarre to me that all three mirrors I use on a day-to-day basis are out of sync, and /only/ on this package 2014-04-17 10:32:28 and since this is a security-related package, call me paranoid, but I just wanted to report it :p 2014-04-17 10:33:26 works here for x86 2014-04-17 10:33:55 works for my x86_64 too 2014-04-17 10:34:26 fabled, did you remove the package first? it will work if the package is already installed 2014-04-17 10:34:26 you are on the new musl builds, right? 2014-04-17 10:34:31 yes indeed I am 2014-04-17 10:34:43 quite enjoying the proper unicode support 2014-04-17 10:35:35 yeah. me too. and that certain stuff no longer crashes. 2014-04-17 10:35:37 like vlc 2014-04-17 10:36:08 well in case this was a temporary glitch, let me nuke this chroot and build it again 2014-04-17 10:48:18 nope, I can confirm it is still broken 2014-04-17 10:48:29 ca-certificates, and abuild both show "short write" and "BAD signature" 2014-04-17 10:50:32 sure that you have enough disk space? 2014-04-17 10:50:59 "BAD signature" implies that it's wrong identity hash 2014-04-17 10:51:13 mismatch with APKINDEX and the actual package file 2014-04-17 10:51:53 awilfox, can get the sha1sums of the ca-certificates.apk and APKINDEX from the mirror ? 2014-04-17 10:52:23 my chroot is: /dev/mapper/vg_pwyll-vm 30G 17G 12G 60% /opt/virtual 2014-04-17 10:54:26 and also "apk policy ca-certificates" 2014-04-17 10:58:32 how /do/ I determine the sha1sum of an apk? it doesn't match, but I tried "sha1sum busybox-1.22.1-r3.apk" and compared to the index and it didn't match either 2014-04-17 11:00:33 the sha1sum in index is not over the full file 2014-04-17 11:05:05 hmm 2014-04-17 11:05:19 http://slexy.org/view/s2MiF3Ywev 2014-04-17 11:05:21 fabled, ^ 2014-04-17 11:05:29 fabled, I can't figure out how to sha1sum but the sizes do not match 2014-04-17 11:05:42 they are short by a few octets 2014-04-17 11:06:00 busybox included as a "good" package 2014-04-17 11:06:13 that is, one I know works, and verifies with the S:line in the index 2014-04-17 11:06:40 -rw-r--r-- 1 tteras tteras 171330 Apr 17 10:52 ca-certificates-20140325-r0.apk 2014-04-17 11:06:51 on my x86_64 2014-04-17 11:07:00 is the file in apk cache? 2014-04-17 11:07:06 I just downloaded it... 2014-04-17 11:07:14 awilcox on pwyll [pts/20 Thu 17 5:52] /tmp: curl -O http://dl-3.alpinelinux.org/alpine/edge/main/x86_64/ca-certificates-20140325-r0.apk 2014-04-17 11:07:32 I receive this file, that is 9 octets short.. 2014-04-17 11:07:45 what mirror do you use, fabled 2014-04-17 11:07:51 nl.a.o 2014-04-17 11:07:58 yes, dl-2 seems to have mismatching size 2014-04-17 11:08:19 also dl-3 2014-04-17 11:08:32 nl.a.o is ok 2014-04-17 11:08:46 ah okay 2014-04-17 11:09:13 i wonder if something went wrong in sync. but its weird, since it's bad on multiple merrors 2014-04-17 11:09:26 abuild and sqlite-libs are also bad 2014-04-17 11:09:31 if that helps 2014-04-17 11:09:34 both short a few bytes 2014-04-17 11:09:44 on both dl-3 and dl-2 2014-04-17 11:13:43 ncopa, who runs dl-[23] ? 2014-04-17 11:14:47 jbilyk? 2014-04-17 11:15:07 i think its jbilyk and/or nangel 2014-04-17 11:15:24 dl-3 == wiki 2014-04-17 11:15:47 i think it's nangel running that 2014-04-17 11:23:07 going to sleep for a bit, if you need any further things re: this ca-certificates issue I will be around in a few hours 2014-04-17 12:37:49 i help maintain dl-3 2014-04-17 12:37:53 ncopa / fabled 2014-04-17 12:38:55 jbilyk, some packages in edge / x86_64 seem to have mismatching file size / checksum 2014-04-17 12:39:14 re-running sync job 2014-04-17 12:39:15 wondering if it's disk / rsync / 2014-04-17 12:39:29 fabled: can you retry with dl-3 2014-04-17 12:39:48 still wrong sizes 2014-04-17 12:39:53 where do you sync from? 2014-04-17 12:39:57 nl.a.o ? 2014-04-17 12:40:18 example file that is wrong is http://dl-3.alpinelinux.org/alpine/edge/main/x86_64/ca-certificates-20140325-r0.apk 2014-04-17 12:40:37 it gives 171321 bytes; instead of 171330 2014-04-17 12:40:41 disk full? 2014-04-17 12:40:47 fabled: grabbing from rsync.a.o 2014-04-17 12:40:52 which I think is nl, yeah 2014-04-17 12:41:00 yeah 2014-04-17 12:41:05 nl gives right file out 2014-04-17 12:41:13 so either rsync is broke, or something funky is going on 2014-04-17 12:41:27 maybe it does not do timestamps correct? 2014-04-17 12:41:39 i tried to apk verify the packages 2014-04-17 12:41:46 but the keys are not there, in v2.4 2014-04-17 12:43:03 dal-a3:/var/www/wiki.alpinelinux.org/htdocs/alpine/edge/main/x86_64# apk verify ca-certificates-20140325-r0.apk 2014-04-17 12:43:03 ca-certificates-20140325-r0.apk: 0 - OK 2014-04-17 12:43:13 fabled: ^^ 2014-04-17 12:43:31 i purged lots of packages 2014-04-17 12:43:35 to generate a rebuild 2014-04-17 12:43:40 fabled: I guess I could force rsync to check all files against nl 2014-04-17 12:43:41 it probably still has the olds 2014-04-17 12:43:44 that'd be a pain 2014-04-17 12:43:53 difference of packages http://sprunge.us/aYYM 2014-04-17 12:44:14 so it's a rebuild 2014-04-17 12:44:24 but why 2014-04-17 12:44:34 i purged packages in x86 2014-04-17 12:44:42 to do rebuild with pie gcc 2014-04-17 12:44:51 relro 2014-04-17 12:44:54 this is with x86_64 2014-04-17 12:44:58 hum 2014-04-17 12:45:06 that makes no sense... 2014-04-17 12:45:21 could be but in buildrepo 2014-04-17 12:45:33 that purges files when it shouldnt 2014-04-17 12:45:37 there's 6 seconds difference 2014-04-17 12:45:39 in build time 2014-04-17 12:45:47 two builers running same time? 2014-04-17 12:45:59 hm 2014-04-17 12:46:02 that could be too 2014-04-17 12:46:03 but more importantly, why is not rsync fixing it 2014-04-17 12:46:16 that probably a rsync flag 2014-04-17 12:46:28 jbilyk, what rsync flags you have? 2014-04-17 12:46:30 from the dl-3 script: 2014-04-17 12:46:31 /usr/bin/rsync -prua \ 2014-04-17 12:46:31 $exclude \ 2014-04-17 12:46:31 --delete \ 2014-04-17 12:46:31 --timeout=600 \ 2014-04-17 12:46:31 --delay-updates \ 2014-04-17 12:46:33 --delete-after \ 2014-04-17 12:46:35 -v -P \ 2014-04-17 12:46:37 "$src" "$dest" 2014-04-17 12:46:55 hm 2014-04-17 12:46:56 or 2014-04-17 12:47:14 maybe its the uclibc packages that are still there? 2014-04-17 12:47:21 oh 2014-04-17 12:47:25 -u = skip files that are newer on the receiver 2014-04-17 12:47:47 it could be that it was caused by the diskfull wi had 2014-04-17 12:48:02 i think dl-2 got full disk 2014-04-17 12:48:05 at some point 2014-04-17 12:48:22 jbilyk, can you remove -u and try again? 2014-04-17 18:32:57 is there any difference in security between aliasing a 2nd ip on an interface - or using 2 interfaces for internal / external networks ? 2014-04-17 20:18:19 hi 2014-04-17 20:18:25 how can I check, if my system runs on musl? 2014-04-17 20:24:30 shafire, apk info -e musl 2014-04-17 20:24:45 shafire, if it prints "musl" it is installed, if it does not then it is not 2014-04-17 20:27:29 :-) ok 2014-04-17 20:27:30 thaks 2014-04-17 20:27:31 thanks 2014-04-17 20:28:16 glad to help 2014-04-17 20:29:02 I am trying to get docker running on alpine 2014-04-17 20:32:48 ERROR: musl-utils-0.9.14-r6: trying to overwrite sbin/ldconfig owned by uclibc-utils-0.9.33.2-r27. 2014-04-17 20:32:53 can I switch fully to musl? 2014-04-17 21:16:22 I downloaded the edge iso from yesterday 2014-04-17 21:16:33 got some kernel panics on start 2014-04-17 21:16:52 after 3 reboots, it worked ... :S 2014-04-17 21:32:32 where can I find the libdevmapper.h? 2014-04-19 12:26:46 any of you guys run intel nuc, gigabyte brix or similar? i have a small request for information. 2014-04-19 12:37:39 aren't those... pretty much a regular box? 2014-04-19 12:38:15 I got a mitx h77 i3 system, I like to think it's in the same league as those, at a slightly larger size 2014-04-19 12:46:04 any idea why i'm not getting new device files in /dev after repartitioning a disk? not even if delete & rescan it via sys? 2014-04-19 12:46:42 restarting mdev adds them, but is that the way? 2014-04-19 15:42:20 ovf: what do you want to know about them? 2014-04-19 15:42:40 clandmeter: setserial -gb /dev/ttyS* 2014-04-19 15:42:49 the output of. 2014-04-19 15:43:53 hmm does it matter which brix? 2014-04-19 15:44:03 the one you have? 2014-04-19 15:44:25 we produce them :p 2014-04-19 15:45:06 ah. then i can pose my question differently :-) 2014-04-19 15:45:49 does any recent one have serial? i know it doesn't have db9 on the front panel, but anything i can connect to? 2014-04-19 15:46:18 we have a debug port on it 2014-04-19 15:46:26 usb debug? 2014-04-19 15:46:28 not sure which interface it has 2014-04-19 15:46:30 no 2014-04-19 15:46:37 diffferent connector 2014-04-19 15:46:56 i think in our rma center they attache a debug card to it. 2014-04-19 15:47:16 I have one brix at home upstairs 2014-04-19 15:47:33 well, if you can get any more information i will... i will... BUY IT! or even two. 2014-04-19 15:47:55 I can get more information for sure. not sure you would want to buy one based on that info ;-) 2014-04-19 15:47:58 basically i'm interested in a small modern system i can debug easily 2014-04-19 15:48:16 debug as in make the os talk to me early 2014-04-19 15:48:58 hi, apk does not have any only-ipv6 support? 2014-04-19 15:49:34 ovf: http://tweakers.net/ext/f/lh2IT47zuA50vzwgy3earutM/full.jpg 2014-04-19 15:49:42 see the white connector? 2014-04-19 15:49:47 on the left 2014-04-19 15:57:34 shafire, what? 2014-04-19 15:57:42 It should work fine on IPv6 only 2014-04-19 15:58:50 clandmeter: yes, seen that. that's what got my hopes up in the first place. but it doesn't provide any indication on what's really going through it. 2014-04-19 15:59:07 clandmeter: so btw if you can run setserial on yours it might be a good first approximation 2014-04-19 16:00:59 awilfox: I tried and it did not work 2014-04-19 16:01:06 I used dl-5.alpinelinux.org 2014-04-19 16:09:05 weird 2014-04-19 16:09:11 nginx is suddenly listening on all interfaces 2014-04-19 16:09:14 when it's not supposed to be 2014-04-19 16:09:25 has anyone else run into this 2014-04-19 17:08:00 clandmeter: oh. and by the way, alpine linux really delivers. 2014-04-19 17:08:42 orly 2014-04-19 17:49:32 "delivers"? 2014-04-19 18:12:58 in this particular case, the deliverable was someone working on exactly the device i was interested in. 2014-04-20 15:06:14 Hi 2014-04-20 15:06:31 I'm having a problem with alpine 2014-04-20 15:06:42 when I mark a message with the "D" flag 2014-04-20 15:06:51 it gets deleted 2014-04-20 15:08:29 it doesn't show the "D" letter but when I enter again to the message index , the message has dessappeared 2014-04-20 15:20:36 pierrot_: while it's quite possible you'll find help here, this channel is for alpine linux the distribution, not the mail client. 2014-04-20 15:24:18 Sorry, I thought that this channel was for the email client 2014-04-20 17:01:03 has anyone not been able to boot after upgrading to musl edge? 2014-04-20 17:01:13 I get thrown into an emergency shell 2014-04-20 18:09:20 I wonder if it's a problem with disk encryption and musl or something 2014-04-20 18:09:22 sigh 2014-04-20 18:09:29 why did I upgrade :( 2014-04-20 18:10:55 Upgraded without any issue here 2014-04-20 18:11:21 I am very satisfied, alpine keeps getting better instead of worse 2014-04-20 18:12:23 are you using disk encryption 2014-04-20 18:39:13 because it's saying it can't mount root 2014-04-20 18:39:24 which is why i'm thinking it's a problem with the encryption 2014-04-20 18:46:27 willing to pay for someone with a solution 2014-04-20 18:46:31 I have no clue how to fix this 2014-04-21 00:49:53 willing to pay for musl edge boot help 2014-04-21 01:10:02 100$/min 2014-04-21 18:11:45 did I ever mention how much I love race conditions? 2014-04-21 18:12:34 http://f.cl.ly/items/1G062X2l2M2f2B2Y2922/arf3352.png 2014-04-21 18:12:40 the answer is: I do not love them at all 2014-04-21 18:59:24 try race conditioner 2014-04-21 18:59:30 but watch out for race conditioning 2014-04-21 19:01:08 what the UNHOLY HELL 2014-04-21 19:01:15 mack:/var/lib/tftpboot# cd .. 2014-04-21 19:01:19 mac:/var/lib/tftpboot/fs# 2014-04-21 19:32:30 ACTION rawrs at TheWilfox 2014-04-21 19:32:40 arf :o 2014-04-21 19:32:44 :3 2014-04-21 19:32:52 ACTION petpets Diftraku, how're you 2014-04-21 19:33:11 great, starting as a summer intern at Tieto tomorrow 2014-04-21 19:33:25 http://www.tieto.com/ 2014-04-21 19:33:57 oooh. what'll you be doing there? 2014-04-21 19:34:07 don't really know yet xD 2014-04-21 19:34:14 data centre stuff 2014-04-21 19:34:26 racking stuff I guess 2014-04-21 19:34:36 aah, fun fun 2014-04-21 19:34:50 the team I'll be in handles all the installation of servers 2014-04-21 19:34:55 hope you enjoy it! 2014-04-21 19:35:09 it'll be awesome no doubt :3 2014-04-21 19:38:42 TheWilfox: y u no on institute for retarded chatters, atheme axis of evil? 2014-04-21 19:39:05 Diftraku: I resigned from there amicably 2014-04-21 19:39:25 Diftraku: found the post-merger network to be insufferable and intolerable, unfortunately 2014-04-21 19:39:49 better than merging with j4ckoff 2014-04-21 19:39:58 Diftraku: http://foxkit.us/goodbye-sb.txt 2014-04-21 19:40:02 omg lol, that dude 2014-04-21 19:40:13 he infested interlinked (another network I'm on) 2014-04-21 19:40:16 " With my own homemade PHP chanserv" 2014-04-21 19:40:31 he was k:lined for PM spamming every oper 2014-04-21 19:40:44 and I do mean every oper. I don't grok that kid in the slightest 2014-04-21 19:43:31 yup 2014-04-21 19:47:50 ACTION rawrs at TheWilfox 2014-04-21 19:48:02 ACTION yips at Diftraku! 2014-04-21 19:49:59 ACTION nests on TheWilfox's lap 2014-04-21 19:50:49 ACTION petpets Diftraku 2014-04-21 19:51:04 rawrr! 2014-04-21 22:21:03 fully-functional Alpine install with Python 3.3, a bunch of random utils, and some custom libs: 12 MB gzipped 2014-04-21 22:21:06 god 2014-04-21 22:21:11 I love alpine so much sometimes 2014-04-21 22:21:17 it's very possibly smaller than fucking /netbsd/ 2014-04-22 01:59:30 willing to pay for alpine musl broken boot help 2014-04-22 02:00:02 ... still? 2014-04-22 02:00:13 yeah 2014-04-22 02:00:17 I have no idea how to fix it man 2014-04-22 02:00:24 I have no clue what is going on 2014-04-22 02:01:01 I charge 215$/hr and require remote root access and all your passwords 2014-04-22 02:01:10 lol 2014-04-22 02:01:27 for that much I'd rather just wipe it and do everything over 2014-04-22 02:01:35 and hope it doesn't break again 2014-04-22 02:01:43 hey, you underestimate me. I have most problems fixed in 6-8 hours. 2014-04-22 02:01:51 lol 2014-04-22 02:02:29 n.b. the problems I specialise in typically end up requiring kgdb and/or null-modem cables. 2014-04-22 02:02:34 (I'm sadly serious) 2014-04-22 02:02:57 whatever my problem is 2014-04-22 02:03:02 february I had a really tricky issue with the DMAR support in xen 4.3 2014-04-22 02:03:15 is probably just some stupid silly command or something that musl broke 2014-04-22 02:03:25 where i would have to wipe everything and not upgrade to musl 2014-04-22 02:03:26 xen 4.3 doesn't support kgdb, either, and I didn't have a USB debugging cable so I had to do it all with sleep()s and kprintf()s 2014-04-22 02:04:26 but yeah 2014-04-22 02:04:29 this is super annoying 2014-04-22 02:04:36 because as long as it doesn't work I can't move everything over to it 2014-04-22 02:04:43 so I'm stuck paying for something I can't use 2014-04-22 02:05:02 why did I upgrade to musl :'( 2014-04-22 02:14:15 and the guy most likely to know the answer 2014-04-22 02:14:19 hasn't been around in several days 2014-04-22 02:14:23 sigh 2014-04-22 06:37:55 Mp5shooter well do you have console access to it, and the part when it boots? 2014-04-22 06:48:28 ? 2014-04-22 06:50:04 krjn well i saw two times his msg about asking help, but non of the issue, or i dont have that log 2014-04-22 06:56:23 Where can I find the build scripts, configuration files, patches etc. for the Alpine toolchain? 2014-04-22 07:02:22 krjn i used this apk add alpine-sdk http://wiki.alpinelinux.org/wiki/Setup_your_system_and_account_for_building_packages 2014-04-22 07:05:07 crow: thanks, but will this not just download the binary toolchain? 2014-04-22 07:12:34 krjn well you can check APKBUILD for toolchain git clone git://dev.alpinelinux.org/aports 2014-04-22 07:13:22 krjn http://git.alpinelinux.org/cgit/aports/ 2014-04-22 07:22:44 www.libressl.org 2014-04-22 07:26:50 krjn if you did not found it here is APKBUILD and inside depends, then check teir APKBUILD .. http://git.alpinelinux.org/cgit/aports/tree/main/alpine-sdk/APKBUILD 2014-04-22 07:32:26 crow: i was just afk, i will look at the links now 2014-04-22 07:41:35 crow: thanks, everything is in the gcc catalog 2014-04-22 07:45:33 krjn good, if i my ask, what you are searching? 2014-04-22 07:48:29 crow: how a hardened uclibc toolchain is built from scratch, first i was searching after sources to the Gentoo hardened toolchain, but these could not be found 2014-04-22 08:08:30 krjn alpinelinux is switching to musl (edge is based on musl) 2014-04-22 08:18:43 crow: i did not think musl supported relro, but maybe this is not a problem 2014-04-22 08:20:31 krjn just to let you know, that alpinelinux is going musl support, that you if can help you start to work on that. 2014-04-22 08:24:18 crow: maybe i will try to setup a machine with Alpine edge to test musl, i normally always run the stable version 2014-04-22 08:28:04 crow: i now see that latest release of musl supports relro protection 2014-04-22 08:35:10 krjn alpine last stable will stay on glibc. from alpinelinux 3.0* it will be based on musl. 2014-04-22 08:35:33 at least i read that here, i am in no way connected to alpine development.. just using this nice distro 2014-04-22 08:40:58 crow: i think you mean uclibc instead of glibc. i hope Alpine with musl will be as good as Alpine with uclibc 2014-04-22 08:45:39 i think alpine with musl have potensial to be better than alpine with uclibc 2014-04-22 08:54:53 by looking at the musl comparison table, there seems to be a lot of benefits over uclibc 2014-04-22 08:56:12 but i not experienced any negative about uclibc in Alpine Linux 2014-04-22 08:57:40 there are a few bugs that we wont fix with uclibc 2014-04-22 08:57:47 cornercases 2014-04-22 08:57:53 threading cornercases 2014-04-22 08:58:07 for example, vlc can hang once in a while on shutdown with uclibc 2014-04-22 08:58:12 and its non-trivial to fix 2014-04-22 08:58:16 with musl it works 2014-04-22 09:01:04 is all hardening techniques used by Alpine supported by musl now? 2014-04-22 09:05:19 krjn, yes. we implemented RELRO to musl 2014-04-22 09:05:35 -D_FORTIFY_SOURCE is not done, but i think we never had it turned on by default 2014-04-22 09:05:43 and that's likely broken in uclibc too 2014-04-22 09:11:21 fabled: can you also compile musl with stack smashing protection now? 2014-04-22 09:37:51 krjn: i believe ssp is enabled by default in alpine/musl yes 2014-04-22 09:41:37 musl seems more and more interesting. I first learned about it in 2012 while building an embedded system, but i did not think it was ready at the time 2014-04-22 09:46:36 it wasnt :) 2014-04-22 11:22:06 I'm guessing there's no interest in systemd? :) 2014-04-22 11:33:44 hehehe ;) my take is that the alpine devs know how to write solid scripts, so no use case for a new init system in a more obscure implementation :) 2014-04-22 11:42:29 well... systemd brings things that openrc scripts can't 2014-04-22 11:49:30 i.e.? 2014-04-22 11:53:12 reliably tracking forked processes for a service 2014-04-22 11:54:02 complex storage topologies 2014-04-22 11:54:06 pcarrier_: there is no way we will do systemd before kdbus is accepted in mainline kernel 2014-04-22 11:54:51 also, systemd is glibc only 2014-04-22 11:55:55 my biggest worry with systemd is the narsissistic management 2014-04-22 11:56:32 please, definitely, never 2014-04-22 11:57:24 sounds reasonable, thanks 2014-04-22 11:57:47 at least there are no current plans for systemd 2014-04-22 11:57:52 most likely never 2014-04-22 11:58:01 (but i dont want promise that either...) 2014-04-22 11:58:22 one can have systemd on arch, similar enough, ... 2014-04-22 11:58:35 was a reasonable way to build it, would you have objections to add it to pkgs and keeping .service files in other packages? 2014-04-22 11:58:44 dunnot yet 2014-04-22 11:59:02 well, maybe is someone else wanted to maintain it 2014-04-22 11:59:06 i have no interest in it 2014-04-22 11:59:16 i have no interest in maintaining it 2014-04-22 11:59:52 but i would like to have some supervision tool, that is compatible with systemds services files 2014-04-22 12:00:12 but no current plant there either 2014-04-22 12:01:03 if someone wants to sync 100s of hours into something he could natively tie in pacemaker or something 2014-04-22 12:01:13 and fix a lot of bugs there 2014-04-22 12:01:45 I'm growingly concerned about alternatives, given at this point all "major" desktop linux distros basically migrated or intend to 2014-04-22 12:02:25 they also migrated to dbus or consolkit and ditch it now 2014-04-22 12:02:46 it's questionable if you don't just waste some time of your life helping them with their toy 2014-04-22 12:02:51 pcarrier_: yes, it worries me too 2014-04-22 12:03:47 I quite like the admin experience it provides, but obviously that's only one part of the puzzle :) 2014-04-22 12:05:15 pcarrier_: i, just yesterday, had this netctl thing it comes with. turned on ipv6 stateless autoconf. activated that profile and, bazoom, lost networking. because it tried to set a sysctl that wasn't there. bad blindly written junior dev code. no error handling. and the unability to set the stateless autoconfig disabled the v4 interfaces, too. 2014-04-22 12:05:27 IF the code was better, it'd be great 2014-04-22 12:05:53 but it's just the same shit as rhel's systemV init, which was also written by some noob who had never seen a state machine in his life 2014-04-22 12:06:07 just taking the next crap they design won't advance anything :(( 2014-04-22 12:06:37 wow, having worked with the maintainers of systemv-init @ redhat I'll refrain from commenting 2014-04-22 12:06:41 at least it would be good to wait till they decided about network config mechanisms 2014-04-22 12:07:01 i have to rewrite a lot of this stuff for some vessel control project 2014-04-22 12:07:13 and the guys doing air traffic control did the same 2014-04-22 12:07:25 but yeah sorry it's getting a bit too close in that case 2014-04-22 12:11:26 Is there a way to change Creation Date of an existing file? 2014-04-22 12:21:58 crow: ctime? 2014-04-22 12:22:12 crow: mode or owner change will udpate the ctime 2014-04-22 12:25:36 fs debuggers can set it too i think, if it's supposed to be a certain value (whatever that would be good for!) 2014-04-22 12:28:45 maybe he wants to cover his trails... 2014-04-22 12:36:45 crow: no more crashes? 2014-04-22 12:37:19 uggedal hmm no ctime command on esx host :( 2014-04-22 12:38:33 clandmeter actually i dont want to cover trails, there is an Nagios check, which looks for only for 4 Files, and i have copied one new file, and is reported in check, and its warning. File copied in same tree. 2014-04-22 12:38:56 clandmeter no as matter of fact no, still using same oscam.debug, didnt have time to switch to lxc oscam on edge-musl 2014-04-22 12:39:02 is planed for this week 2014-04-22 13:00:38 I see that docker was moved into the main aports tree last week. What needs to be done before a binary package can be added to the edge repository? 2014-04-22 13:22:57 mwcampbell, if it's in main/ a binary package should be automatically built 2014-04-22 13:23:03 if it's not there, it means there has been a build failure 2014-04-22 13:30:17 errm 2014-04-22 13:30:31 is there a reason why mod_mpm_event has disappeared from the latest version of the apache2 package? 2014-04-22 13:32:06 hl, build failure. the edge builders currently do not report error; but just silently delete failed packages 2014-04-22 13:32:19 we should fix that now that musl is migrated to be ege 2014-04-22 13:32:20 ncopa, ^^^ 2014-04-22 13:32:22 no, this isn't a separate package 2014-04-22 13:32:54 oh 2014-04-22 13:33:12 upstream removed it? or it was intentionally disabled from configuration? 2014-04-22 13:33:51 Upgrading apache2 (2.4.6-r4 -> 2.4.9-r0) 2014-04-22 13:34:02 This is not edge either, and the APKBUILD still specifies 'all' MPMs 2014-04-22 13:34:15 if I had to guess, I'd say it'd been built on a machine without some prerequisite for mpm_event 2014-04-22 13:34:41 hl: alpine v2.7? 2014-04-22 13:34:46 yeah. 2014-04-22 13:35:10 hl: could you please file a bug for it so I dont forget it 2014-04-22 13:35:14 ncopa: OK 2014-04-22 19:59:38 > air traffic control > systemd 2014-04-22 19:59:44 suddenly I am afraid of flying again ;) 2014-04-22 20:04:16 i think you read it the wrong way :)) 2014-04-22 20:04:42 imagine if any of that networkmanager avahi systemd stuff were not disabled on such systemds 2014-04-22 20:04:51 minus 'd' lol 2014-04-22 20:05:01 lol 2014-04-22 20:05:21 "something happened, we'll now turn off your interfaces" errr.... "what!?!" 2014-04-22 20:52:48 TheWilfox, considering a LOT of ATC systems date from the 70's... 2014-04-22 20:52:49 :p 2014-04-22 20:54:38 I still trust CP/M-80 more than lennartp 2014-04-22 20:54:40 j/s 2014-04-22 20:55:57 I dunno 2014-04-22 20:56:08 the tech used to control $200 million planes is less advanced than that used to control $20,000 cars 2014-04-22 20:56:33 well, "less advanced" also means "simpler", sometimes 2014-04-22 20:56:45 depends. hopefully they didn't use something like SNOBOL or F77 2014-04-22 20:57:34 they probably did. 2014-04-22 20:57:59 idk, Ada was pretty popular in aviation. it was designed for government anyway 2014-04-23 00:06:59 have been playing with alpine for the last couple of hours and love it 2014-04-23 00:07:55 yeah alpine is great 2014-04-23 04:42:10 moinmoin 2014-04-23 06:56:54 is there a tarred rootfs available? 2014-04-23 07:03:14 ovf, you can create in 1 2 cmd's 2014-04-23 07:03:23 -1 2014-04-23 07:04:02 i didnt finish my coffee yet so spelling mistakes may occur 2014-04-23 07:04:29 yes, using http://wiki.alpinelinux.org/wiki/Installing_Alpine_Linux_in_a_chroot . was just wondering if you provided one (even archlinux finally came around to do this) 2014-04-23 07:04:48 no we don't 2014-04-23 07:05:27 one can use apk static and use the root option 2014-04-23 07:05:52 its so simple, no need to nurse others. 2014-04-23 07:12:16 ok, understood. 2014-04-23 07:15:58 clandmeter: any update on brix, btw? 2014-04-23 07:16:13 oh forgot 2014-04-23 07:16:19 let me call rma 2014-04-23 07:16:36 well call, let me walk cause english is not their speciality. 2014-04-23 07:38:22 ovf: they dont have info regarding that connector. they have a debug card to attache to it. they will ask more info, but dont expect much. 2014-04-23 07:40:01 clandmeter: thank you very much 2014-04-23 07:40:39 when i get some time (at home) ill check serial on my own brix 2014-04-23 09:13:47 ovf: you can extract files from .iso with 7zip 2014-04-23 09:14:21 oh, rootfs 2014-04-23 09:22:27 :) 2014-04-23 09:22:49 ncopa: coffee? 2014-04-23 09:23:58 i had mine already... :) 2014-04-23 09:24:15 but apparently it was not strong enough :) 2014-04-23 09:35:04 ovf: i have good and bad news 2014-04-23 09:35:24 bad news, its only for debug purpos, so no info will be provided 2014-04-23 09:36:13 good news, i just got a new one in my hands which will have support for com ports (they have additional ports on the pcb). 2014-04-23 09:55:59 clandmeter: awesome! (the good part). the new one isn't in production yet, right? 2014-04-23 09:56:52 ovf: its the Bay Trail version 2014-04-23 09:58:01 probaly end of next month 2014-04-23 09:58:42 it will have some kind of io addon card. but not for endusers i think. 2014-04-23 09:59:12 if you buy 1k you can get them ;-) 2014-04-23 10:03:42 well, it has the headers regardless 2014-04-23 10:04:26 i would be more content with haswell over bay trail, but sometimes you just have to settle for what's there i guess 2014-04-23 15:39:39 Hi 2014-04-23 15:39:48 I get this error: 2014-04-23 15:39:59 apk add curl 2014-04-23 15:40:04 ERROR: unsatisfiable constraints: 2014-04-23 15:40:12 ca-certificates (missing): 2014-04-23 15:40:20 required by: curl-7.36.0-r0[ca-certificates] 2014-04-23 15:40:40 anybody knows? 2014-04-23 15:46:19 I just tried upgrading to edge/main and now apk can't find symbols in libcrypto. I just saw above that there is a static build of apk so I'm going to try fixing with that but wanted to report here. I can open a bug if someone would like for me to. 2014-04-23 15:50:36 seems force is needed for musl upgrade as well 2014-04-23 15:50:39 xaocon: i think you *must* use the static apk for the update 2014-04-23 15:50:48 didn't try yet 2014-04-23 15:55:54 well it turns out you do right now but it isn't mentioned at all in the upgrade to edge docs 2014-04-23 16:06:36 Confirmed: ca-certificates package exists into /2.7/main/ but not into /edge/main/ repositories 2014-04-23 18:39:51 was the gold linker left out of binutils just to keep the package small? 2014-04-23 19:47:24 xaocon: it should be implemented soon (from what i read) 2014-04-23 19:55:25 to the person who had issues with ca-certificates, it's fixed now 2014-04-23 19:55:38 not sure who it was (scrollback is erased) but I did see someone having issues earlier 2014-04-23 19:58:37 It was AmatCoder, though he's gone now. 2014-04-23 19:59:33 aah 2014-04-24 03:06:50 why have you made it so hard to submit patches now 2014-04-24 03:07:10 I was around when it was just "file a ticket in redmine", well now my redmine creds don't work and the wiki says I have to email shit in to a list 2014-04-24 03:07:33 of course, I can't even make smtp work, but why do I have to email fucking patches to a ML, why can't I just use redmine like before 2014-04-24 03:07:36 it's a simple one line change 2014-04-24 03:07:49 linux-vanilla doesn't have "xz" in makedepends 2014-04-24 03:08:40 http://slexy.org/view/s2O4pGhcby 2014-04-24 03:08:56 this is such a fucking simple patch why is it so hard to contribute this patch to you to improve the distro I like 2014-04-24 05:29:03 TheWilfox: busybox has unxz which is used in abuild so there should be no need to depend on xz 2014-04-24 05:29:17 uggedal: yes, but busybox doesn't have 'xz' 2014-04-24 05:29:21 and 'xz' is used in the build 2014-04-24 05:29:24 as I said in the patch 2014-04-24 05:29:41 The kernel won't build without xz; it fails with: 2014-04-24 05:29:44 linux-3.12/scripts/xz_wrap.sh: exec: line 23: xz: not found 2014-04-24 05:53:02 TheWilfox: I overlooked the patch url. Seems like a bug. The official builders probably have xz installed for performance reasons 2014-04-24 05:54:44 morning 2014-04-24 05:55:13 TheWilfox: doesnt that depend on the config, if you need xz or not? 2014-04-24 05:55:21 ncopa, I am not sure 2014-04-24 05:55:30 also I am sorry for being a little grumpy there earlier 2014-04-24 05:55:43 had to fight with it for hours because the 'xz: not found' line was always off scrollback 2014-04-24 05:55:48 it was just a bunch of 'Make: error' 2014-04-24 05:55:53 so it made me a bit upset 2014-04-24 05:55:57 iirc you can configure kernel image to compress with xz 2014-04-24 05:56:03 or initramfs image 2014-04-24 05:57:23 kernelconfig.x86:CONFIG_HAVE_KERNEL_XZ=y 2014-04-24 05:57:34 kernelconfig.x86_64:CONFIG_HAVE_KERNEL_XZ=y 2014-04-24 05:58:30 hm 2014-04-24 05:58:34 must be something else 2014-04-24 05:58:44 i dont have xz installed here 2014-04-24 05:58:53 linux-vanilla builds fine for you? 2014-04-24 05:58:56 and i dont think the builders has it either 2014-04-24 05:58:59 apparently... 2014-04-24 05:59:05 do you have custom kernel config? 2014-04-24 05:59:06 that is quite odd 2014-04-24 05:59:12 I disabled some modules 2014-04-24 05:59:15 i'm rebuilding now 2014-04-24 05:59:15 ah 2014-04-24 05:59:17 I don't believe I changed any compression sessions 2014-04-24 05:59:24 *settings 2014-04-24 06:00:01 it could be that enabling some things will disable the xz support 2014-04-24 06:00:21 so it could be that the official kernel does not actually have xz support 2014-04-24 06:00:37 and when you disabled the things that disables xz, you got xz enabled 2014-04-24 06:00:48 oh god, lol 2014-04-24 06:00:58 just guessing... :) 2014-04-24 06:01:00 okay I'm sorry for bothering you all and wasting your time then 2014-04-24 06:01:08 ha 2014-04-24 06:01:19 it was just very annoying (took hours to figure out) and thought I might save someone else the hassle :P 2014-04-24 06:01:28 http://sprunge.us/gQLP 2014-04-24 06:01:33 that is the *running* kernel 2014-04-24 06:01:42 # CONFIG_KERNEL_XZ is not set 2014-04-24 06:02:03 CONFIG_KERNEL_XZ=y 2014-04-24 06:02:08 yes indeedy 2014-04-24 06:02:14 something I disabled must have flicked that on. 2014-04-24 06:02:35 but interesting though 2014-04-24 06:02:46 since i think the intention is that is should be enabled? 2014-04-24 06:03:19 no 2014-04-24 06:03:25 CONFIG_KERNEL_GZIP=y 2014-04-24 06:03:42 its the kernel image 2014-04-24 06:03:53 the way kernel image is compressed 2014-04-24 06:04:06 we use gz 2014-04-24 06:05:26 TheWilfox: we shoudl fix your creds for redmine 2014-04-24 06:05:37 i'm ok that you attach patches there if thats easier for you 2014-04-24 06:06:09 it is 2014-04-24 06:06:14 I have not had working email in many years 2014-04-24 06:06:21 can receive sometimes, but not send 2014-04-24 06:06:23 what was your login username? 2014-04-24 06:06:35 it should have been 'awilcox' 2014-04-24 06:06:44 my email address (on redmine) is awilcox@wilcox-tech.com 2014-04-24 06:06:51 found you 2014-04-24 06:07:23 oh wow thank you! 2014-04-24 06:07:29 login works now, thanks so much 2014-04-24 06:07:51 was my account locked out or such? 2014-04-24 06:08:20 dunno really 2014-04-24 06:08:27 i didnt really do anything... :) 2014-04-24 06:08:37 lol, weird 2014-04-24 06:08:41 maybe was a temporary glitch 2014-04-24 06:08:51 i was looking on how to reset password for you, but didnt find it 2014-04-24 06:08:57 yes sounds like a temporary glitch 2014-04-24 06:09:13 maybe NSA was doing hardware upgrade or something... :) 2014-04-24 06:09:21 hehehe 2014-04-24 06:10:03 i'm not sure i will apply the patch though 2014-04-24 06:10:22 i dont tihnk we want have all deps for all possible config combinations 2014-04-24 06:10:42 and i think this is config dependant 2014-04-24 06:11:01 that said, adding xz isnt a big deal so if you have strong feelings for it, then i'll add it 2014-04-24 06:11:56 nah it is okay 2014-04-24 06:12:04 I didn't realise my module stuff changed other things 2014-04-24 06:12:16 this is more a fault with menuconfig than alpine now ;) 2014-04-24 06:12:40 ok. good. I'm glad you figured it out :) 2014-04-24 06:13:20 me too, thanks 2014-04-24 06:44:43 Hi all! 2014-04-24 06:44:57 I have a AL 2.3.6 HDD install 2014-04-24 06:45:24 Is there anything known to prevent me to just update /etc/apk/repositories and start using 2.7 instead? 2014-04-24 06:45:29 (note that it's a HDD install) 2014-04-24 06:47:37 According to http://wiki.alpinelinux.org/wiki/Upgrading_Alpine#Upgrading_an_Alpine_Linux_Hard-disk_installation it shouldn't be a problem 2014-04-24 06:48:01 I just wanted to make sure there's no inconsistency between 2.3.x and 2.7.x 2014-04-24 06:50:59 And... I'm mostly thinking of apk-tools and other base functions (I don't expect you to promise that all other applications continue working as they used to) :-) 2014-04-24 07:00:33 It wasn't that straight forward as I hoped - so I canceled the upgrade 2014-04-24 07:00:48 I bring up a new box instead - just to be sure the current box don't break 2014-04-24 07:59:31 I'd do upgrade to v2.4 first 2014-04-24 07:59:36 2.3 is unsupported 2014-04-24 09:17:39 Good Morning all! 2014-04-24 09:27:18 not sure if anyone particularly cares, but following the randomness discussion days ago on here, I've finally written up a wiki article: http://wiki.alpinelinux.org/wiki/Entropy_and_randomness 2014-04-24 09:27:50 Haven't had as much time for testing various things as I'd hoped 2014-04-24 09:33:19 ty 2014-04-24 09:33:36 np, hope it comes in useful to somebody somewhere! 2014-04-24 09:58:58 ginjachris: nice knowhow-collection, thx! 2014-04-24 11:28:30 nice! 2014-04-24 11:29:25 yeah 2014-04-26 22:49:44 Hi. 2014-04-27 17:21:02 what do folks use for graphical network management on Alpine? 2014-04-27 17:47:37 Alpine has a webinterface for doing various configuration. Run 'setup-acf' and you would have it up and running quickly. 2014-04-27 18:07:40 mhavela: more specifically I'm looking for an answer like "wicd" or "NetworkManager/nm-applet", I'm sick of running /etc/init.d/wpa_supplicant restart ; udhcpc -i wlan0 2014-04-27 18:08:25 Oh. I see. 2014-04-27 18:09:05 I agree - it would be interesting to hear how others do (personally I created a script that I called 'setup-wireless' which does what I need regarding wireless) 2014-04-27 18:09:19 Hopefully someone else has some good comments 2014-04-28 06:25:22 duck1: I use dhcpcd/dhcpcd-gtk 2014-04-28 06:25:34 and manually mangage wifi connections 2014-04-28 06:25:43 i think we have networkmanager in testing 2014-04-28 06:25:47 i havent used that though 2014-04-28 09:14:33 Hello... 2014-04-28 09:15:36 Is there a way to tell 'crond' not to log everything? I have to run a job every minute and it's kinda annoying to have the log full of these entries, so I'd like to NOT log for that particular job. 2014-04-28 09:25:17 danci1973: cant you let crond output to a different file? 2014-04-28 09:25:40 clandmeter: Yes, of course you can... 2014-04-28 09:26:38 clandmeter: In fact, my current solution is to have 'crond' log to '/dev/null' (by adding -L /dev/null startup option), but that's not quite what I'd want either. 2014-04-28 09:28:15 so you want per cron job log options? 2014-04-28 09:37:05 danci1973: did you try set -l 20? 2014-04-28 09:39:50 clandmeter: Ideally there would be some way to let crond know not to log when a particular cron job is run (maybe pre-pending a ! in front of the time spec)... I would like to keep normal logging for normal jobs that occur once a day, or a couple of times at most... 2014-04-28 09:40:45 clandmeter: But for now, redirecting all logs to /dev/null is 'good enough'. 2014-04-28 09:40:57 danci1973: use a different cron daemon that provides those opitons 2014-04-28 09:41:19 /dev/null sounds bad 2014-04-28 09:41:34 atleast you would want to know when an error happens 2014-04-28 09:44:32 clandmeter: Agreed, but in this case it's no worse than having /var/log/messages cluttered by1440 identical entires every day. 2014-04-28 09:47:21 did you try -l setting? 2014-04-28 09:47:26 i think 20 will only display errors 2014-04-28 09:53:50 clandmeter: I'll try it out. 2014-04-28 14:17:13 hi 2014-04-28 14:18:30 will http://wiki.alpinelinux.org/wiki/How_to_setup_a_wireless_access_point ever be updated says 2012 2014-04-28 14:20:44 hm 2014-04-28 14:20:45 its a wiki 2014-04-28 14:20:57 i havent used hostapd 2014-04-28 14:21:13 so someone who has will have to contribute 2014-04-28 14:47:54 zxd: seems like the original author didn't fully test: http://wiki.alpinelinux.org/w/index.php?title=How_to_setup_a_wireless_access_point&diff=4709&oldid=4704 2014-04-28 14:48:10 so ymmv 2014-04-28 16:18:32 ncopa: thanks for the pointer to dhcpd-gtk, I'd really like to see wicd or connman in addition to networkmanager just so there's some options for graphical wifi control + dhcp/net config options. Personally, I've found networkmanager to be poorly designed with bugs creeping out that takes them a year+ to resolve for the last 5-7 years. 2014-04-28 23:35:07 Hi there, is it know that using the default x86_64 image with the default setup-alpine, a system is created that does not boot? 2014-04-28 23:40:54 i think i solved, it but i probably unfold a bug, i'll try to find the bugtracker ;) 2014-04-29 07:00:30 o/ 2014-04-29 08:41:15 does someone here now, a package name for updating ip address within no-ip 2014-04-29 08:47:32 maybe ez-ipupdate can do it? 2014-04-29 09:05:01 ncopa let me check 2014-04-29 09:05:54 is there an officiel howto for upgrade stable -> edge-musl 2014-04-29 09:06:54 http://lists.alpinelinux.org/alpine-devel/3637.html 2014-04-29 09:07:05 crow: mailing list and www i think 2014-04-29 09:07:19 http://alpinelinux.org/edge-musl 2014-04-29 09:07:21 http://kaniini.dereferenced.org/2014/03/23/upgrading-to-alpine-musl.html 2014-04-29 09:07:59 ok thanks 2014-04-29 09:19:09 ncopa clandmeter this errors are safe to ignore and do reboot? http://sprunge.us/UbPW 2014-04-29 09:20:05 crow: try run an 'apk fix' first 2014-04-29 09:21:06 ncopa ok see the end of paste http://sprunge.us/iagF 2014-04-29 09:21:32 you can also try remove the uclbic files its complaining for in /etc/mkinitfs/base.files 2014-04-29 09:22:06 i believe it should be ok 2014-04-29 09:22:10 well with apk fix it did reinstalled musl so i guess it got them remove them 2014-04-29 09:22:23 yes, its above 2014-04-29 09:22:29 (13/39) Purging uclibc-utils (0.9.33.2-r27) 2014-04-29 09:22:37 ncopa and the /etc/mkinitfs/base.files is empty 2014-04-29 09:22:44 oh 2014-04-29 09:22:52 hang on a sec then... 2014-04-29 09:23:04 sorry 2014-04-29 09:23:12 i mean /etc/mkinitfs/features.d/base.files 2014-04-29 09:24:06 compare it with the /etc/mkinitfs/features/bas.files.apk-new 2014-04-29 09:24:15 base.files.apk-new 2014-04-29 09:24:31 http://sprunge.us/dCRK 2014-04-29 09:24:36 ok just as i copyed it to sprunge http://sprunge.us/dCRK 2014-04-29 09:25:25 i think you can just remove the lines starting with /lib/ 2014-04-29 09:25:29 http://sprunge.us/Fhei 2014-04-29 09:25:41 yes seems so 2014-04-29 09:25:48 looks good 2014-04-29 09:25:58 then apk fix mkinitfs 2014-04-29 09:26:01 or i can just rename base.files and rename base.files.apk-new to base.files 2014-04-29 09:26:17 should be good enough, unless you have local modifications to those 2014-04-29 09:26:34 no just clean install 2014-04-29 09:28:02 then it should be ok to replace wiht .apk-new 2014-04-29 09:28:16 other than that, everything looked good to me 2014-04-29 09:29:06 uclibc got removed which means you have nothing that hold uclibc back 2014-04-29 09:29:56 whole log again http://sprunge.us/QNFC 2014-04-29 09:31:28 looks good to me 2014-04-29 09:31:35 nothing that indicates that it should not work 2014-04-29 09:31:42 nothing that i can see at least 2014-04-29 09:31:54 yea doing reboot 2014-04-29 09:48:13 booted fine, 2014-04-29 09:48:41 ez-ipupdate does not support no-ip and seems not maintend anymore. Some other would be updatedd but seems not in alpinelinux 2014-04-29 09:52:25 one could be also ddns http://wiki.openwrt.org/doc/howto/ddns.client?s[]=no&s[]=ip 2014-04-29 09:53:48 hmm is this commint in alpinelinux https://github.com/wimpunk/ddclient/pull/1/files?short_path=04c6e90&unchanged=collapsed (google find it) 2014-04-29 09:59:09 ah seems they just added init.d service in ddclient source 2014-04-29 09:59:34 and you need perl :| 2014-04-29 09:59:38 https://duckdns.org/install.jsp#linux-cron 2014-04-29 09:59:50 this is free and needs curl ;-) 2014-04-29 10:00:08 https://github.com/wimpunk/ddclient 2014-04-29 10:00:14 clandmeter yea int need perll 2014-04-29 10:01:36 clandmeter but that is only for duckdns.org 2014-04-29 10:02:05 sure 2014-04-29 10:02:18 thats why its better to use duckdns :) 2014-04-29 10:03:06 not an option 2014-04-29 10:03:32 but seems these ddns-script are used from openwrt to http://wiki.openwrt.org/doc/howto/ddns.client?s[]=no&s[]=ip 2014-04-29 10:14:30 this seems to use only glibc 2014-04-29 10:14:30 https://projects.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/noip 2014-04-29 15:57:17 asides from storage domUs being a PITA why t**f**k does shitty qemu default to writeback mode 2014-04-29 15:57:20 qemu-system-i386: -drive file=/dev/zvol/zdata/freebsd1,if=ide,index=3,media=disk,format=raw,cache=writeback: could not open disk image /dev/zvol/zdata/freebsd1: No such file or directory 2014-04-29 15:57:42 finally explains why i had <1GB/s read but >1GB/s write on freebsd 2014-04-29 15:58:30 i wish i could sue people for every bit of data lost due to defaults tuned to dev's laptop 2014-04-29 16:00:22 darkfader: I guess you are using ZFS on Linux? 2014-04-29 16:00:26 no 2014-04-29 16:00:33 in a freebsd domU 2014-04-29 16:01:28 then you won't be able to launch HVM guests using storage on the driver domain 2014-04-29 16:01:53 *drops jaw* 2014-04-29 16:02:10 darkfader: unless you use a qemu stubdomain :) 2014-04-29 16:02:46 well i'm trying to make one undocumented unreliable feature work & documented, adding another one won't raise chances 2014-04-29 16:03:10 what's making hvm-from-hvm impossible? 2014-04-29 16:03:20 or actually maybe hvm-from-pvhvm 2014-04-29 16:03:52 don't even know if the freebsd vm really works in pvhvm since there's no place on earth documenting how to tell if pvh actually is being used 2014-04-29 16:04:26 it's not because of the hvm-from-hvm combination, it's because Qemu requires access to the physical disk in order to boot the HVM domain, which you don't have if the disk is on another domain 2014-04-29 16:04:35 ah so qemu is the problem 2014-04-29 16:04:45 so if i used phy instead of raw? 2014-04-29 16:05:36 FreeBSD will automagically choose the best possible mode for you, if you look at the dmesg you will see things like xen_et, xenbusb_front being attached 2014-04-29 16:06:20 doesn't help at all... is that the only two or are there 3 more i don't know of :) 2014-04-29 16:06:30 (just to make you see the problem) 2014-04-29 16:06:56 in general you should not worry, FreeBSD should choose the best possible mode for you 2014-04-29 16:07:32 i'm out of words 2014-04-29 16:08:09 i.e. there's a pv clock driver for hvm, so yes, that will autoenable 2014-04-29 16:08:17 but it'll do that either way 2014-04-29 16:08:28 well, PVHVM is not a "mode" like PV or HVM, is more or less like a collection of PV hacks that you can use to make HVM faster 2014-04-29 16:08:35 yes 2014-04-29 16:08:44 and HVM is fucking slow compared to pv 2014-04-29 16:08:53 so it makes a difference if 5 hacks are active or 3 2014-04-29 16:09:25 mmm, in almost all workloads PVHVM is faster than PV 2014-04-29 16:09:43 i said HVM is slow 2014-04-29 16:09:48 not PVHVM is slow 2014-04-29 16:10:09 and if i had PVHVM mode I would like it to be able to use all tweaks it has 2014-04-29 16:10:27 and not find later that I missed one bios setting or freebsd patch that turned off half of them 2014-04-29 16:11:20 I mean, you did a lot of work to make it work with freebsd, wouldn't you like people to be able to fully use it? 2014-04-29 16:11:55 people are fully using it, if it's not used it's because it's not available 2014-04-29 16:12:00 I've always had the PV drivers compiled in on freebsd vms so it's not like I can easily tell 2014-04-29 16:12:30 are you saying this because on FreeBSD 10 the XENHVM kernel config is gone? 2014-04-29 16:12:38 no 2014-04-29 16:13:32 i'm the person who pushed colin to goddamn put the 3 relevant lines from xenhvm into generic 2014-04-29 16:17:12 colin had the idea to put this all in a module, but at the end we managed to just put it in the GENERIC config and get done with it 2014-04-29 16:17:44 yeah it's the way it should be 2014-04-29 16:18:03 that should have been done in 2007 :) 2014-04-29 16:18:22 yes, not sure why it wasn't done before (fear to broken generic stuff I guess) 2014-04-29 16:18:39 mostly because too high goals like dom0 support 2014-04-29 16:19:08 but yeah, because of that: if you don't know the N(tweaks) that turn on with PVH you can't find out if it worked. 2014-04-29 16:19:22 I hope it does, but i can't see a difference if i have pvh=1 in the config or not 2014-04-29 16:19:31 i know the hardware is the right one for it 2014-04-29 16:19:53 (pvh='1' to be 100% correct) 2014-04-29 16:20:31 adding pvh=1 to a (PV)HVM config won't do anything, you need to create a PV domain 2014-04-29 16:21:06 so as long as you have builder='hvm' it won't do pvhvm? 2014-04-29 16:21:29 heyho got a lil problem here I try to install alpine to my hd which is mounted on /mnt (ext4) but it kinda does nothing but create the etc dir. -v does nothing after outputting boot filesystem: ext4 2014-04-29 16:21:34 no, PVHVM is one thing, PVH is another completely different 2014-04-29 16:21:50 blkid is running and a grep for ^/dev/sda2 2014-04-29 16:22:11 darkfader: for PVH you have to follow: http://wiki.xen.org/wiki/FreeBSD_PVH 2014-04-29 16:22:13 i think pvh was called pvhvm in the original slides? 2014-04-29 16:22:16 i used that 2014-04-29 16:22:54 and i refuse to call hvm "installed by someone who's not too stupid to install pv drivers" pvhvm 2014-04-29 16:22:58 what a mess 2014-04-29 16:23:14 if you look at the link you gave 2014-04-29 16:23:34 does one still need to use that forked kernel? 2014-04-29 16:24:30 yes, I still haven't fully commited the remaining patches 2014-04-29 16:24:47 would like to do it this week, let's see how that turns out 2014-04-29 16:25:01 so the big xen blog announcement was just about adding the drivers from xenhvm to generic? 2014-04-29 16:25:17 i'm not the only one who fell for that 2014-04-29 16:25:49 (and that is why any documentation should have _some_ info on how to verify it worked 2014-04-29 16:25:55 mmm, what a mess, the blog post was about PVHVM, the wiki page is about PVH, which is a new virtualization mode that appeared in Xen 4.4 2014-04-29 16:26:41 and which many people waited for since the initial announcement 2014-04-29 16:26:56 those people who had their pv drivers working for ages 2014-04-29 16:27:07 pvhvm as a term is annoying 2014-04-29 16:27:21 i mean, do you call a windows vm with drivers installed pvhvm too? 2014-04-29 16:27:48 no, this would be a HVM domain with PV drivers, windows cannot do PVHVM 2014-04-29 16:28:26 darkfader: http://wiki.xen.org/wiki/Virtualization_Spectrum should help I guess :) 2014-04-29 16:29:01 yeah pvh is the one i'm after 2014-04-29 16:29:28 ever since it was announced (and back then it was actually a lot slower than pv, but i hope that's sorted) 2014-04-29 16:30:29 but that page is interesting. 2014-04-29 16:30:42 just to make it clear, PVH has (more or less) the same performance as PVHVM, the nice things about PVH is that it can run as Dom0, which PVHVM cannot 2014-04-29 16:31:03 oh, ok 2014-04-29 16:32:20 so it'll just need less integration in linux to run as dom0? 2014-04-29 16:32:31 to make lkml happier? 2014-04-29 16:32:32 :) 2014-04-29 16:33:30 yes, good guess, PVH is much less intrusive than PV 2014-04-29 16:34:33 I even have a FreeBSD PVH Dom0, and it doesn't require many changes 2014-04-29 16:35:43 cool 2014-04-29 16:36:53 i'm yet to see what's best for combining storage and dom0 OS... the 4 ssds that gave 2GB/s in linux do 1GB/s on freebsd 2014-04-29 16:37:43 i'm still confused if i have it all ideally configured now 2014-04-29 16:37:51 i'll reread the backlog a few times tonight 2014-04-29 16:38:34 darkfader: I guess you are doing a pci-passthrough to the FreeBSD DomU right? 2014-04-29 16:38:34 the page you linked is kinda funny though 2014-04-29 16:38:37 "But there is one PV guest that can never be run in PVHVM mode, and that is domain 0. Because having a domain 0 with the current Linux drivers will always be necessary, it will always be necessary to have a PV mode in the Linux kernel." 2014-04-29 16:38:58 there's millions of linux PV domUs 2014-04-29 16:39:11 that work just fine 2014-04-29 16:39:24 yet, most of the dev time goes everywhere else :) 2014-04-29 16:39:29 royger: yup 2014-04-29 16:39:46 right now i get the disks back via iscsi 2014-04-29 16:40:04 and the performance is from inside of the FreeBSD DomU with the passthrought, or from the guests that are using those disks? 2014-04-29 16:40:44 I see, so you are not using FreeBSD blkback 2014-04-29 16:42:24 the disk in the domU, but even if i boot freebsd natively 2014-04-29 16:42:47 (the freebsd passthrough domU i mean) 2014-04-29 16:43:04 the only difference i found so far is that freebsd only sets 128 queue depth 2014-04-29 16:43:13 on Linux I have something like 970 2014-04-29 16:43:43 darkfader: I'm afraid I'm not of much help with this, never really tweaked all those internals 2014-04-29 16:43:54 np :) 2014-04-29 16:44:12 there might be some sysctl tuneable to change this... 2014-04-29 16:44:13 right now the only goal is to make the storage domain work 2014-04-29 16:44:54 anyway I have to leave, I will read the scrollback later if you have anything else :) 2014-04-29 16:45:13 ty a lot! 2014-04-29 16:45:17 and get home well :) 2014-04-29 16:45:41 I'm already home ;) leaving to buy dinner 2014-04-30 07:24:47 Hi! I did a 'apk upgrade -U' on a 2.7.x box (I think it was 2.7.4 but I am not 100% sure). 2014-04-30 07:25:00 After the upgrade, the color output on rc disapeared 2014-04-30 07:25:12 Does anybody know if this is intentional? 2014-04-30 07:26:04 ('rc-status' used to display the '[started]' or '[crashed]' in green/red) 2014-04-30 07:26:15 I still have color output on my ssh sessions though 2014-04-30 07:26:35 This is *not* a high prio task - as it only is all about 'looks' :-) 2014-04-30 07:26:52 'looks' on the terminal 2014-04-30 08:37:03 no, that was not intentional 2014-04-30 08:37:41 i have no idea why it happened either 2014-04-30 09:06:14 here i have colors 2014-04-30 09:06:33 over ssh 2014-04-30 09:09:46 http://abload.de/img/rc-status01lqjwa.png 2014-04-30 09:45:52 Confirmed - I have colors through ssh too